<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Thinuri Wickramarachchi on Medium]]></title>
        <description><![CDATA[Stories by Thinuri Wickramarachchi on Medium]]></description>
        <link>https://medium.com/@thinuridulsini?source=rss-2e65eb35360b------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*gvbf4b9cxHH7D6bcCnqdHw.jpeg</url>
            <title>Stories by Thinuri Wickramarachchi on Medium</title>
            <link>https://medium.com/@thinuridulsini?source=rss-2e65eb35360b------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Fri, 20 Nov 2026 06:29:24 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@thinuridulsini/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[100,000 Vulnerabilities in One Year | Why Critical Bugs Are Dropping Back-to-Back and How We…]]></title>
            <link>https://medium.com/@thinuridulsini/100-000-vulnerabilities-in-one-year-why-critical-bugs-are-dropping-back-to-back-and-how-we-b3c7637ee4d0?source=rss-2e65eb35360b------2</link>
            <guid isPermaLink="false">https://medium.com/p/b3c7637ee4d0</guid>
            <category><![CDATA[programming]]></category>
            <category><![CDATA[java]]></category>
            <category><![CDATA[cybersecurity]]></category>
            <category><![CDATA[devops]]></category>
            <dc:creator><![CDATA[Thinuri Wickramarachchi]]></dc:creator>
            <pubDate>Thu, 19 Nov 2026 02:11:01 GMT</pubDate>
            <atom:updated>2026-11-19T02:11:01.550Z</atom:updated>
            <content:encoded><![CDATA[<h3>100,000 Vulnerabilities in One Year | Why Critical Bugs Are Dropping Back-to-Back and How We Survive It</h3><blockquote>From PHP2Shell to Java4Shell, AI just broke the old rules of software security. Here is why your newsfeed feels like an endless emergency alert and the exact playbook to get ready.</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*zLgjPuW9AQ2vgLpbGcN9lw.png" /></figure><p>If you feel like you barely finished patching one “once-in-a-decade” vulnerability before the next emergency alert hijacked your week, you aren’t imagining things.</p><p>In late October 2026, researchers dropped <strong>PHP2Shell</strong>, an AI-discovered flaw in the core PHP engine threatening 70% of the web. Less than a month later, on November 17, <strong>Java4Shell (CVE-2026–114938)</strong> hit a critical memory corruption bug buried inside OpenJDK’s runtime, uncovered using OpenAI’s Codex in roughly 20 minutes of active prompting for less than $200. Between those two headlines, total disclosed vulnerabilities in 2026 quietly shattered <strong>100,000 CVEs</strong> more than double all of 2025.</p><p>Welcome to the machine speed vulnerability era. The internet isn’t suddenly written worse than it was five years ago; rather, <strong>artificial intelligence has permanently broken the economics of finding, writing, and weaponizing bugs</strong>.</p><p>Here is what is actually driving this relentless wave of security news and how developers, DevOps engineers, and tech leaders can build a system that doesn’t panic every Tuesday.</p><h3>1. AI Is Auditing 30 Years of “Trusted” Core Code</h3><p>For decades, foundational runtimes like OpenJDK, the Linux kernel, and core web engines were protected by sheer complexity. Finding a memory bug inside thousands of lines of handwritten x86–64 assembly or C++ required elite human researchers spending months tracing edge cases.</p><p>Today, AI auditing agents chew through that decades-old technical debt in hours. Anthropic’s Claude Mythos recently surfaced thousands of high-severity flaws including a 27-year-old bug in OpenBSD and a 16-year-old flaw in FFmpeg while Calif’s “centaur” AI workflow cracked OpenJDK’s Base64 assembly in a single afternoon. Even more importantly, Google Threat Intelligence Group data shows AI doesn’t just generate low-level noise: AI-assisted discovery disproportionately uncovers <strong>High-Risk and Remote Code Execution (RCE) flaws</strong>.</p><h3>2. AI Copilots Are Shipping Features and Subtle Bugs 3x Faster</h3><p>While offensive AI is auditing old code, defensive engineering teams are using AI coding assistants to ship pull requests at record speed.</p><p>More code means more surface area. CISA’s 2026 <em>Quality Era</em> framework explicitly warned that automated and AI-enabled development is introducing massive new pressures across the software lifecycle. AI-generated commits frequently nail the “happy path” while subtly skipping authorization checks, edge-case input validation, or proper error handling and because pull requests are three times larger, exhausted human reviewers approve them faster.</p><h3>3. 24/7 Cloud Fuzzing &amp; Aggressive Transparency</h3><p>Alongside LLMs, continuous cloud-scale fuzzing now bombards open-source repositories with millions of malformed inputs around the clock. (In fact, an automated fuzzer independently triggered the Java4Shell crash in just three minutes.) Combined with over 550 authorized CVE Numbering Authorities publishing 300 to 550 records a day, bugs that used to be quietly fixed in minor releases now hit public feeds with formal CVE numbers.</p><h3>4. The “Time-to-Exploit” Window Has Collapsed</h3><p>The scariest shift isn’t the volume of bugs it’s the speed of weaponization. PHP2Shell was actively exploited in the wild within five days of disclosure, and commercial RCE exploits for Java4Shell were built within hours of Oracle’s patch release. Attackers no longer spend weeks reverse-engineering patches; AI helps them diff updates and build working exploit chains overnight.</p><h3>First, Breathe: Why 100,000 CVEs Doesn’t Mean 100,000 Fires</h3><p>Before you cancel your weekend plans, look at the signal-to-noise ratio. While raw CVE volume more than doubled in 2026, actual in-the-wild exploited vulnerabilities added to CISA’s Known Exploited Vulnerabilities (KEV) catalog grew by only about <strong>11%</strong>.</p><p>The vast majority of newly discovered bugs require rare configurations, local access, or complex heap grooming that criminals rarely bother with when unpatched perimeter servers are still sitting wide open. <strong>If your team treats every CVE like a five-alarm fire, you will burn out before the real threat arrives.</strong></p><h3>How We Get Ready</h3><p>You cannot out-type a machine, and you cannot manually patch 300 alerts a day. Surviving back-to-back disclosures requires changing <em>how</em> you prioritize and isolate your infrastructure:</p><p><strong>1. Ditch “Severity-Only” Patching for Risk-Based Triage</strong></p><p>Stop sorting your backlog purely by CVSS score. Prioritize patches using three filters:</p><ul><li><strong>Active Exploitation &amp; Exploitability:</strong> Is it in CISA’s KEV catalog, or does it have a high EPSS (Exploit Prediction Scoring System) score?</li><li><strong>Internet Exposure:</strong> Is the vulnerable service facing the public internet especially identity providers (like Keycloak or PingFederate), CI/CD pipelines (Jenkins, Artifactory), or collaboration hubs (Jira, Confluence)?</li><li><strong>Runtime Reachability:</strong> Does your specific hardware or code path actually execute the flawed function?</li></ul><p><strong>2. Eliminate Your “Hidden Runtime” Blind Spots</strong> Both PHP2Shell and Java4Shell exposed a painful truth: running apt upgrade on your Linux server does not update your containers or bundled apps. Modern apps hide private JDKs, Python interpreters, and Node runtimes deep inside Docker layers and vendor appliances. Maintain automated Software Bills of Materials (SBOMs) and build CI/CD pipelines that can rebuild and redeploy base container images in under an hour.</p><p><strong>3. Assume Breach: Lock Down East-West and Outbound Traffic</strong></p><p>When a zero-day drops before a patch exists, network architecture is your safety net:</p><ul><li><strong>Zero-Trust Microsegmentation:</strong> If an attacker compromises a web container, strict internal firewall rules should stop them from moving laterally to your database or internal Kafka/Elasticsearch clusters.</li><li><strong>Egress Filtering (Outbound Controls):</strong> As security researcher samczsun proved with Java4Shell, a backend worker with <em>zero</em> open inbound ports can still be hacked if it initiates an outbound TLS connection to a malicious server. Restrict outbound internet access from production workloads to explicitly allowlisted domains.</li></ul><p><strong>4. Put AI Guardrails on AI-Generated Code</strong></p><p>If your developers use AI to write code faster, use AI and automated security gates to verify it just as fast. Mandate static analysis (SAST), dependency scanning, and automated fuzzing inside every pull request pipeline, and require heightened human security review whenever code touches authentication, serialization, or binary/Base64 parsing.</p><p><strong>5. Rehearse Safe “Kill Switches” Before an Emergency Hits</strong> When an emergency advisory drops, rushed copy-pasting causes self-inflicted outages like teams globally deploying -XX:UseAVX=2 during Java4Shell and accidentally crashing every ARM64 server in their cloud fleet. Document safe runtime mitigation flags (such as -XX:-UseBASE64Intrinsics), WAF virtual-patching rules, and live process verification commands (jcmd) in your incident runbooks so your team executes muscle memory instead of guesswork.</p><p>The surge of back-to-back vulnerabilities isn’t a temporary storm it is the new baseline of software engineering. AI is finally forcing the tech industry to confront decades of hidden complexity inside the tools we take for granted. The teams that win in this era won’t be the ones who never see a critical CVE; they will be the ones who know exactly what runtimes they are running, isolate their blast radius, and can ship a verified patch before the hype cycle even peaks.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b3c7637ee4d0" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The AI Found It, Then the Attackers Weaponized It | Inside the Apache HTTP/2 Crisis…]]></title>
            <link>https://medium.com/@thinuridulsini/the-ai-found-it-then-the-attackers-weaponized-it-inside-the-apache-http-2-crisis-2aad7e697409?source=rss-2e65eb35360b------2</link>
            <guid isPermaLink="false">https://medium.com/p/2aad7e697409</guid>
            <dc:creator><![CDATA[Thinuri Wickramarachchi]]></dc:creator>
            <pubDate>Wed, 18 Nov 2026 15:36:26 GMT</pubDate>
            <atom:updated>2026-11-18T15:36:26.837Z</atom:updated>
            <content:encoded><![CDATA[<h3>The AI Found It, Then the Attackers Weaponized It | Inside the Apache HTTP/2 Crisis (CVE-2026–119618)</h3><h3>First Java4Shell, now Apache. AI is auditing the internet at machine speed, and CVE-2026–119618 is the latest critical unauthenticated RCE being actively targeted in the wild. Here’s how we got here and how to lock it down.</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*S9t3hmV_zQoofehHuUvYSQ.png" /></figure><p>Just when you thought you had your UseBASE64Intrinsics Java flags under control, the internet’s favorite web server started flashing red.</p><p>Only days after the chaotic disclosure of <strong>Java4Shell</strong>, cyber security agencies issued emergency warnings that <strong>CVE-2026–119618</strong> a critical unauthenticated Remote Code Execution (RCE) vulnerability in the ubiquitous <strong>Apache HTTP Server</strong> is now seeing active exploitation in the wild. A public proof-of-concept (PoC) is circulatiing, and automated scanner traffic is surging.</p><p>Like its sibling Java4Shell, the core story isn’t just about the code flaw; <strong>it is how the bug was found.</strong></p><p>CVE-2026–119618 was uncovered during automated research audits by <strong>Anthropic’s OSS Scanner</strong> (using AI) and <strong>Google’s Cloud Vulnerability Research</strong>. What once would have taken human analysts months of tracing through asynchronous code was identified by AI agents in minutes.</p><p>Now, unpatched web servers globally are caught in a race against AI-assisted exploitation.</p><p>Here is what CVE-2026–119618 actually is, why it exists, how we got here, and the exact steps to triage your systems before the weekend.</p><h3>What Is It? The mod_http2 Race Condition</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*tFQO1SSJC5keRD-3dIPRLQ.png" /></figure><p>The vulnerability lives inside <strong>mod_http2</strong>, the Apache module responsible for handling HTTP/2 traffic. To understand the bug, we have to look at how modern web servers scale.</p><p>Unlike HTTP/1.1, which opens a new TCP connection for every single file (images, scripts, styles), HTTP/2 utilizes <strong>multiplexing</strong>. It allows multiple requests and responses to fly back and forth simultaneously over a single connection, handled concurrently across different processor threads.</p><p>The point of failure is a thread-safety bug (CWE-362) inside Apache’s <strong>assign_headers()</strong> function. When multiple concurrent HTTP/2 requests overload the server, two or more threads can try to write, modify, and process headers at the exact same millisecond without proper synchronization.</p><ol><li><strong>The Concurrency Gap:</strong> An unauthenticated attacker sends a custom stream of highly multiplexed HTTP/2 headers designed to confuse the server’s threading logic.</li><li><strong>The Heap Corruption:</strong> The race condition causes the memory pointers of one thread to overwrite the heap memory spaces of another thread.</li><li><strong>The Unauthenticated Takeover:</strong> By carefully manipulating the payload, this heap corruption allows the remote attacker to bypass authentication boundaries entirely and run arbitrary commands directly as the Apache process (www-data or apache).</li></ol><h3>Why Is This Happening Now?</h3><p>This isn’t bad luck; it’s a symptom of a larger shift in cybersecurity. Over the last decade, web architecture rushed to support HTTP/2 to make page loads faster, while the foundational C-based code underneath (like Apache’s httpd) had to be heavily adapted to handle massive thread concurrency.</p><p>Asynchronous, multi-threaded C/C++ code is notorious for thread safety and synchronization bugs, which are notoriously difficult for standard human code audits to trace because they only trigger under specific “race conditions” on busy servers.</p><p>But in late 2026, those hidden race conditions have become low-hanging fruit for AI agents. AI doesn’t suffer from “trace fatigue”; it can simulate and visualize complex thread concurrency states across millions of codebase permutations to see exactly where memory bounds fail.</p><p>Defenders are using AI to discover these vulnerabilities in Apache and Java but once the patches drop, adversaries are immediately using AI to help reverse-engineer the patches and target unpatched, exposed internet assets.</p><h3>How to Find Out If You Are At Risk</h3><p>To be vulnerable to CVE-2026–119618, a server must meet <strong>all three</strong> of the following conditions:</p><ol><li><strong>Vulnerable Version:</strong> Running Apache HTTP Server versions <strong>2.4.55 through 2.4.70</strong> (or standard Git mod_h2 versions 2.0.8 through 2.0.42).</li><li><strong>HTTP/2 Enabled:</strong> You must be actively using the mod_http2 module with HTTP/2 enabled (h2 or h2c protocols in configuration). <em>If you only serve HTTP/1.1, you are safe.</em></li><li><strong>External Reachability:</strong> If your Apache server is directly facing the internet without a reverse proxy (such as a Content Delivery Network or an HTTP load balancer that translates traffic back down to HTTP/1.1 before it hits Apache), you are completely exposed.</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*2mH6zkLslrzSjOjifUC2og.png" /></figure><h4>Why your automated OS package checks might lie to you:</h4><p>Many distribution maintainers backport security fixes without changing the main version number. For example, on an Ubuntu or Debian server, running apache2 -v might show 2.4.52 (which seems older and safe), but you might still have backported vulnerability updates if the packages were updated within the last few days. Conversely, you might see 2.4.62 and think you are safe, despite being directly within the 2.4.55 to 2.4.70 vulnerability window.</p><p>The most reliable way to check is to inspect your live configuration for HTTP/2 use:</p><pre># Check if the mod_http2 module is loaded<br>apachectl -M | grep http2</pre><h3>How to Remediate</h3><p><strong>1. Apply the Official Patch (The Best Fix)</strong> The ultimate fix is updating Apache HTTP Server to version <strong>2.4.71</strong> (or applying your OS distribution’s patched security build released after November 11)</p><ul><li><strong>Debian/Ubuntu:</strong></li><li>sudo apt update &amp;&amp; sudo apt install --only-upgrade apache2 sudo systemctl restart apache2</li><li><strong>Red Hat / CentOS / Rocky Linux:</strong></li><li>sudo dnf upgrade httpd sudo systemctl restart httpd</li></ul><p><strong>2. Turn Off HTTP/2 (The Instant Mitigation)</strong> If you cannot patch immediately due to change management rules or vendor lock-in, you can completely neutralize the vulnerability by disabling HTTP/2.</p><p>Open your Apache configurations (usually /etc/apache2/apache2.conf, /etc/httpd/conf/httpd.conf, or specific vhost config files) and ensure the following:</p><ul><li>Comment out or remove h2 and h2c from your Protocols directive, forcing Apache to stick to HTTP/1.1:</li></ul><p><strong>Apache</strong></p><ul><li># Restrict to HTTP/1.1 (and optionally HTTP/1.0) Protocols http/1.1</li><li>Disable the HTTP/2 module entirely if it isn’t strictly required:</li></ul><p><strong>On Ubuntu/Debian:</strong></p><ul><li>sudo a2dismod http2 sudo systemctl restart apache2</li></ul><p><strong>On RHEL/CentOS:</strong> Comment out the LoadModule http2_module modules/mod_http2.so line in your configuration and restart the service.</p><p><strong>3. Use Load Balancers / Reverse Proxies (The Isolation Rule)</strong> Ensure your Apache server sits behind a modern reverse proxy or load balancer (like Cloudflare, AWS ALB, NGINX) that handles the external HTTP/2 connection and translates it down to HTTP/1.1 before it reaches the backend Apache instances. This breaks the exploit’s multiplexed race condition before it ever hits your servers.</p><h3>Final Thoughts</h3><p>Java4Shell and CVE-2026–119618 are proof that the paradigm of cyber defense has shifted. “Secure” protocols and libraries used by millions of internet touchpoints are being automated, audited, and bypassed in real-time.</p><p>Teams that rely on quarterly maintenance schedules will consistently find themselves exposed. Start automating your patch deployments, map out where HTTP/2 is actually needed, and build containerized pipelines so you can rebuild, test, and ship your web server fleets in under an hour. Because in this AI security era, the gap between finding a bug and exploiting it has vanished.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2aad7e697409" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[A $200 AI Prompt Just Broke Java’s “Safe” Memory Model | Java4Shell (CVE-2026–114938)]]></title>
            <link>https://medium.com/@thinuridulsini/a-200-ai-prompt-just-broke-javas-safe-memory-model-java4shell-cve-2026-114938-d85f92bc0643?source=rss-2e65eb35360b------2</link>
            <guid isPermaLink="false">https://medium.com/p/d85f92bc0643</guid>
            <category><![CDATA[vulnerability]]></category>
            <category><![CDATA[java4shell]]></category>
            <category><![CDATA[openjdk]]></category>
            <dc:creator><![CDATA[Thinuri Wickramarachchi]]></dc:creator>
            <pubDate>Wed, 18 Nov 2026 12:36:49 GMT</pubDate>
            <atom:updated>2026-11-18T12:36:49.226Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*HkHbxzrA7BslCjylbHvPYg.png" /></figure><blockquote>Inside Java4Shell (CVE-2026–114938), the dormant GitHub pull request that hid in plain sight, and how one stray = character compromises Keycloak, Jira, and Jenkins.</blockquote><p>OpenJDK is one of the most scrutinized pieces of software on earth. Millions of enterprise servers rely on its promise of automatic memory management to prevent the exact kind of buffer overflows that plague C and C++ applications.</p><p>Yet on November 17, 2026, Oracle, CISA, and the OpenJDK Vulnerability Group issued emergency warnings for <strong>CVE-2026–114938</strong> dubbed <strong>Java4Shell</strong>. Even wilder than the bug itself is how it surfaced: security researchers at Calif used AI-assisted auditing (Codex) to uncover the flaw in about an hour, spending roughly 20 minutes of active prompting and less than $200 in token costs.</p><p>While Java4Shell has not yet seen confirmed exploitation in the wild, researchers have already demonstrated pre-authentication Remote Code Execution (RCE) against default installations of <strong>Keycloak, Jira, Confluence, Jenkins, JFrog Artifactory, and PingFederate</strong>.</p><p>Here is what actually happened under the hood, why the internet is buzzing about a companion leak called <strong>JavaBleed</strong>, and what every developer and sysadmin needs to do right now.</p><h3>What Is Java4Shell?</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*52goV7EQqj0qV71sGvMlAQ.png" /></figure><p>Unlike 2021’s infamous Log4Shell which lived inside a third-party logging library <strong>Java4Shell lives directly inside the Java runtime itself</strong>.</p><p>Specifically, it hides inside java.util.Base64.Decoder. Whenever an application converts Base64 text back into binary data like reading an HTTP Authorization: Basic login header, parsing a SAML single-sign-on token, or handling a file upload it calls this decoder.</p><p>Because servers have to decode your login header <em>before</em> they know who you are, an attacker doesn’t need a username or password to reach this code. Send a specially malformed Base64 string over the network, and you trigger a heap buffer overflow right inside the Java Virtual Machine (JVM).</p><h3>How It Happened: Subtracting 1 From 0 at Hardware Speed</h3><p>To make Java fast, OpenJDK’s Just-In-Time (JIT) compiler doesn’t always run standard Java bytecode. Once a method runs frequently enough, the JVM swaps it out for handwritten, hardware-accelerated x86–64 assembly code called an <strong>intrinsic</strong>.</p><p>On modern processors supporting <strong>AVX-512 VBMI</strong> instructions (Intel Ice Lake/Sapphire Rapids and AMD Zen 4 or newer), the Base64 intrinsic processes incoming text in high-speed 64-byte chunks and passes any leftover characters at the end to a cleanup routine.</p><p>Here is where basic math breaks down:</p><ol><li><strong>The Malformed Input:</strong> An attacker sends a string whose length isn’t divisible by 4 and ends with a stray padding character such as 128 A characters followed by a single = (129 bytes total).</li><li><strong>The Buffer Allocation:</strong> Java allocates a 98-byte destination array. The fast loop decodes the first 128 As into 96 bytes, leaving just <strong>2 bytes</strong> of safe room at the end of the buffer.</li><li><strong>The Integer Underflow:</strong> The cleanup routine receives the remaining 1-byte = fragment. It calculates the output size as 0, and then subtracts 1 to account for the trailing = without checking if the result goes negative. The output size underflows to <strong>-1</strong>.</li><li><strong>The 62-Byte Spill:</strong> The CPU passes -1 into an AVX-512 bitmask instruction (BZHI). Reading the lowest byte (255), the processor interprets -1 as “write all 64 bytes.” Instead of writing zero bytes, it blasts <strong>64 bytes</strong> into a 2-byte space overwriting <strong>62 bytes</strong> of neighboring objects on the Java heap.</li></ol><p>Java immediately realizes the padding is invalid and throws an IllegalArgumentException. <strong>By then, it’s too late.</strong> The neighboring memory is already corrupted.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*tf4v9tCobiewOM0n4CXRNw.png" /></figure><h4>Turning a Crash Into Takeover: Enter “JavaBleed”</h4><p>On its own, the overflow writes a repeating byte pattern (69 a6 9a) rather than arbitrary attacker code, which usually just crashes the JVM. However, Calif researchers used “Heap Feng Shui” to carefully position objects in memory so those 62 spilled bytes overwrite the length field of an adjacent array.</p><p>Once an attacker enlarges a Java array’s length field, they gain arbitrary read/write access to JVM memory. Calif dubbed the resulting memory-disclosure technique <strong>JavaBleed</strong> allowing an attacker to scrape passwords, private keys, and memory addresses (defeating ASLR) straight out of RAM before hijacking full control of the server.</p><p><em>(In an ironic twist, an open-source contributor had actually spotted an AVX-512 Base64 overflow and submitted a public pull request to OpenJDK on September 9, 2026 complete with a sample payload. It sat unmerged for eight weeks until an automated inactivity bot closed it in early November, days before the vulnerability leaked ahead of Oracle’s scheduled patch.)</em></p><h3>Real-World Impact: Who Is Actually at Risk?</h3><p>To be exploitable, a system needs <strong>all three</strong> of these ingredients:</p><ul><li><strong>Vulnerable Java Version:</strong> OpenJDK (and downstream builds like Oracle Java SE, Corretto, Temurin, Zulu, and SapMachine) versions <strong>17.0.6 through 17.0.21</strong>, <strong>21 through 21.0.13</strong>, <strong>25 through 25.0.5</strong>, and <strong>27 through 27.0.1</strong> (plus Oracle’s specialized Java 8 Enterprise Performance Pack 8u511b33-perf). Standard Java 8 and Java 11 are <strong>not</strong> affected.</li><li><strong>Specific CPU Hardware:</strong> x86–64 chips with AVX-512 VBMI (Intel Xeon Ice Lake+, 2019–2021 Intel Core Ice/Tiger/Rocket Lake, and AMD EPYC/Ryzen Zen 4+). <strong>ARM64 processors like AWS Graviton and Apple Silicon are completely immune.</strong></li><li><strong>Reachable Decoder:</strong> Untrusted input reaching Base64.getDecoder() or Base64.getUrlDecoder() (getMimeDecoder() is unaffected).</li></ul><p><strong>What researchers have proven so far:</strong></p><ul><li><strong>Pre-Auth RCE:</strong> Calif achieved full, unauthenticated code execution against default setups of <strong>Keycloak, PingFederate, Jira, Confluence, JFrog Artifactory, and Jenkins</strong>, plus post-auth RCE on <strong>Kafka, Elasticsearch, Solr, Gerrit, and Spinnaker</strong>.</li><li><strong>Outbound Client Attacks:</strong> Security researcher samczsun demonstrated that a Java client with zero open inbound ports can still be hacked simply by making an outbound TLS connection to a malicious server.</li><li><strong>Crash vs. RCE Reality Check:</strong> Independent labs at Blackpoint Cyber, Hadrian, and Upwind confirmed how easy denial-of-service is Blackpoint crashed an unauthenticated Spring Boot app in 20 seconds using ~28,300 malformed login requests and read adjacent secrets 30 out of 30 times while noting that reliable RCE in the wild requires heavy, app-specific heap grooming.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*nKRclpDzNrpKvr9PoLrShQ.png" /></figure><h3>How to Fix It (Without Accidentally Taking Down Production)</h3><p><strong>1. Patch Your Java Runtimes (The Real Fix)</strong> Upgrade immediately to the patched .1 security builds released on November 17–18, 2026 such as <strong>Oracle Java SE </strong><strong>17.0.21.1, </strong><strong>21.0.13.1, </strong><strong>25.0.5.1, </strong><strong>27.0.1.1</strong>, <strong>Amazon Corretto </strong><strong>17.0.21.1.2</strong>, or the November <strong>Azul Zulu / SapMachine</strong> updates and restart your services.</p><p><strong>2. Check Your CPUs to Prioritize Triage</strong> On Linux servers, run:</p><pre>grep -om1 -E &#39;avx512[_]?vbmi&#39; /proc/cpuinfo</pre><p>If nothing prints, that specific host cannot trigger the bug today (though you should still patch it in case cloud auto-scaling moves your container to a newer host tomorrow).</p><p><strong>3. Use the Safe Interim Flag (If You Can’t Patch Yet)</strong> If you are waiting on a vendor build (such as Red Hat or Eclipse Temurin) or a maintenance window, add this exact flag to your JVM startup arguments and restart:</p><pre>-XX:-UseBASE64Intrinsics</pre><p><strong>Two critical outage warnings:</strong></p><ul><li><strong>Do NOT use </strong><strong>-XX:UseAVX=2</strong> across mixed fleets ARM64 servers will refuse to boot.</li><li><strong>Do NOT use </strong><strong>-XX:DisableIntrinsic=_base64_decodeBlock alone </strong>Upwind warned that setting it globally via JAVA_TOOL_OPTIONS prevents the JVM from starting at all.</li></ul><p><strong>4. Verify the Running Process</strong> Never trust java -version alone. Confirm the live process actually picked up the mitigation:</p><pre>jcmd &lt;PID&gt; VM.flags -all | grep -F UseBASE64Intrinsics<br># Expected output: bool UseBASE64Intrinsics = false</pre><h3>6 Golden Security Lessons for Beginners</h3><p>If you are newer to backend engineering or cloud security, Java4Shell is a masterclass in how modern vulnerabilities actually work:</p><ul><li><strong>Base64 is encoding, not encryption and it runs before login.</strong> Because Base64 just turns binary bytes into plain text for HTTP headers, attackers can hit your Base64 parser before your authentication code ever runs.</li><li><strong>“Memory-safe” languages still run C and assembly underneath.</strong> Java protects your application code from pointer bugs, but the JVM itself relies on native assembly for speed. When low-level optimizations skip bounds checks, classic memory corruption returns.</li><li><strong>A </strong><strong>try/catch block cannot undo native heap corruption.</strong> Even if your Java code catches IllegalArgumentException, the native assembly routine has already overwritten neighboring memory before throwing that error.</li><li><strong>Updating Linux doesn’t update Docker containers.</strong> Host OS patches won’t touch the private JDKs bundled inside container images or commercial apps like Jira and Keycloak. You must rebuild and redeploy your images.</li><li><strong>Never copy-paste global JVM flags into production untested.</strong> Always verify startup flags on both x86–64 and ARM64 staging nodes before pushing environment variables fleet-wide.</li><li><strong>Firewalls blocking inbound traffic don’t protect outbound clients.</strong> Even internal background workers are exposed if they connect to untrusted external endpoints over TLS.</li></ul><p>As AI-assisted code auditing drops the cost of finding deep compiler bugs below $200, the margin for patching delays is shrinking fast. Knowing every Java runtime hidden inside your containers and verifying live JVM flags instead of assuming is no longer optional hygiene; it is basic survival.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d85f92bc0643" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Inside the November 2026 Microsoft 365 Outage | What Broke, Why Fixes Stalled, and the Hidden…]]></title>
            <link>https://medium.com/@thinuridulsini/inside-the-november-2026-microsoft-365-outage-what-broke-why-fixes-stalled-and-the-hidden-402105a8cc46?source=rss-2e65eb35360b------2</link>
            <guid isPermaLink="false">https://medium.com/p/402105a8cc46</guid>
            <category><![CDATA[microsoft]]></category>
            <category><![CDATA[m365]]></category>
            <category><![CDATA[outage]]></category>
            <category><![CDATA[microsoft-outage]]></category>
            <dc:creator><![CDATA[Thinuri Wickramarachchi]]></dc:creator>
            <pubDate>Fri, 13 Nov 2026 06:11:22 GMT</pubDate>
            <atom:updated>2026-11-13T06:11:22.902Z</atom:updated>
            <content:encoded><![CDATA[<h3>Inside the November 2026 Microsoft 365 Outage | What Broke, Why Fixes Stalled, and the Hidden Cascade</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*0NxhG8PIliPxxxALdze8ZA.png" /></figure><p>On Thursday, November 12, 2026, a localized Exchange Online disruption snowballed within roughly two hours into a global Microsoft 365 outage spanning 74 countries and territories, according to outage tracker Downdetector. Across North and South America, Europe, and into Friday morning across Asia-Pacific, millions of knowledge workers found themselves locked out of Outlook inboxes, struggling with degraded Microsoft 365 tools, watching third-party AI assistants lose access to company files, and improvising workarounds across essential government and healthcare operations. At its early-afternoon U.S. Eastern peak, Downdetector logged up to 138,058 user-submitted reports for Outlook alone, while Microsoft opened two major incident tracks EX1550543 for Exchange Online mailbox access and MO1550821 for broader Microsoft 365 suite connectivity.</p><p>More than twelve hours after the disruption began, full recovery remained elusive. Microsoft explicitly ruled out a sudden surge in incoming traffic and stated it had found no indication of a cyberattack. Instead, engineering updates pointed to an internal cascading availability failure: a subset of servers began failing automated health checks while exhibiting elevated CPU usage, traced specifically to a server process responsible for establishing secure connections. As failing machines dropped out of rotation, surviving servers absorbed the displaced workload, overwhelmed their own secure-connection processes, and dropped offline in a self-reinforcing loop.</p><p>Crucially, as recovery efforts continued into Friday across Asia-Pacific, Microsoft had not yet published a completed root-cause analysis (Root Cause: N/A in its administrative logs) or confirmed a timeline for full restoration, and it has not disclosed an exact count of affected users or organizations. While Microsoft began rolling back a small number of recent configuration and network updates that coincided with the incident’s onset-and Australian Home Affairs Minister Tony Burke noted Microsoft’s local leadership believed an update issue sparked the failure-engineers also warned late Thursday night U.S. time that initial serviceability fixes were not taking as expected.</p><h3>What Actually Broke: From Mailbox Lockouts to Suite-Wide Friction</h3><p>For most users, the outage announced itself shortly after noon Eastern Time on November 12 as a simple inability to load email. Outlook failed across every major connection method web browsers, desktop clients (MAPI), mobile apps (Exchange ActiveSync), and REST APIs leaving inboxes blank and outgoing messages stuck in limbo. Within two hours of its initial Exchange Online warning, however, Microsoft broadened the alert from the Americas to potentially “any region” and acknowledged a second, suite-wide incident (MO1550821) encompassing far more than email.</p><p>By late afternoon in the United States, Microsoft’s administrative advisories listed Exchange Online, Microsoft Teams, SharePoint Online, Microsoft Copilot 365, Microsoft Planner, Microsoft To Do, Power BI (specifically exporting sensitivity-labeled reports to PDF), the Microsoft 365 Apps admin center, and core security and device-management platforms Microsoft Defender XDR, Defender for Office 365, Microsoft Intune, and Microsoft Purview as potentially affected. Some OneDrive and Azure users also reported trouble on Downdetector. Later in the evening, Microsoft clarified that core Teams chat, online meetings, and direct file access in SharePoint remained largely functional, but anything dependent on Exchange Online such as scheduling Teams meetings, creating new chats, or retrieving mailbox-linked context remained severely disrupted.</p><h3>Why It Happened: Anatomy of a Cascading Cloud Failure</h3><p>To understand why a Thursday morning hiccup disabled a hyperscale productivity suite for more than half a day, it helps to trace how Microsoft’s own diagnosis evolved over ten hours of troubleshooting. Initially, early advisories pointed to “elevated load” across infrastructure in multiple regions, leading many IT teams to assume a surge in user traffic was saturating Microsoft’s servers.</p><p>At 2:31 p.m. ET (19:31 UTC), Microsoft corrected that premise. The elevated load was not driven by an influx of user requests, engineers explained, but by a shrinking pool of available machines. Servers were failing automated health checks and becoming unexpectedly unavailable, forcing the remaining infrastructure to shoulder the excess traffic. In distributed cloud systems, automated health probes are designed to protect users by pulling unhealthy machines out of service. When a systemic software or configuration fault causes numerous nodes to fail those checks simultaneously, however, the safeguard turns into a multiplier: each ejected server pushes higher request volume onto its neighbors, spiking CPU utilization until those neighbors trip their own health checks and drop offline as well.</p><p>Minutes later, Microsoft confirmed elevated processor usage on affected servers and targeted the component responsible for processing automated health checks, testing a workaround to bypass it while restarting impacted machines. Nearly six hours into the public response, around 8:48 p.m. EST (01:48 UTC on November 13), Microsoft narrowed the failure mechanism further: the specific server process responsible for establishing secure connections was becoming overwhelmed, causing affected machines to stop processing requests altogether. Independent commentary reported by The Guardian from RMIT University engineering professor Mark Gregory similarly pointed to disruption at the network layer of Microsoft’s infrastructure.</p><p>What originally tipped over that secure-connection bottleneck remains preliminary. Early public updates stated Microsoft had not identified a recent change matching the outage’s start. Yet internal admin-console updates logged at 2:43 p.m. ET (19:43 UTC) noted the health-check failures might be linked to a recent change in the check-processing component. As a precaution, Microsoft paused all deployments and configuration changes. By Thursday evening, engineers identified a small number of recent network and configuration updates coinciding with the outage’s onset and began reversing them to evaluate whether rolling them back accelerated recovery. Even so, Microsoft’s official service health dashboards continued listing the definitive root cause as pending while diagnostics continued.</p><h3>Why Recovery Kept Stalling</h3><p>In routine cloud disruptions, isolating damaged clusters, draining traffic to healthy zones, and rebooting servers clears backlogs within an hour or two. Throughout Thursday afternoon, Microsoft appeared to be making steady headway using exactly those steps shifting traffic to alternate routes, temporarily limiting which types of traffic could reach strained infrastructure, and initiating broad server restarts. By 5:47–6:00 p.m. ET, crowdsourced Downdetector complaints in the U.S. had dropped roughly 75 percent from their peak, falling from 138,058 Outlook reports around 1:00 p.m. ET to about 34,490, while Microsoft 365 suite reports fell to nearly 13,000.</p><p>Then stabilization hit a wall. Microsoft cautioned that servers returned to service could become unresponsive again once they began receiving live traffic, noting that engineers were still analyzing diagnostic data to pinpoint what was triggering that relapse. At 10:19 p.m. EST (03:19 UTC on November 13), as Asia-Pacific offices moved through their Friday workday, Microsoft issued a sobering status update: changes applied to improve serviceability were “not taking as expected,” prompting engineers to pursue alternate workstreams to increase capacity in the face of “sustained traffic management imbalances”. Public incident advisories mirrored by Japan’s Tokai National Higher Education and Research System (THERS) also showed Microsoft evaluating whether isolating traffic within individual regions could allow regional clusters to recover independently.</p><h3>End-to-End Timeline: From Quiet Degradation to Global Disruption</h3><p>Because the outage spanned Thursday in the Americas and Friday across Europe, Asia, and Oceania, tracking the incident in Coordinated Universal Time (UTC) alongside U.S. Eastern Time (ET / EST, UTC-5) clarifies how fast the failure propagated. Notably, incident logs compiled by technical monitors point to a 97-minute gap between Microsoft opening the Exchange Online ticket inside admin consoles and acknowledging the disruption publicly on X.</p><p>Timestamp (UTC)U.S. Eastern TimeKey DevelopmentSource ContextNov 12, 14:45–15:53Nov 12, 9:45–10:53 a.m.Separate Teams outage (TM1550488) hits North and South America and is marked resolved; no confirmed link to later failure.Nov 12, 16:17Nov 12, 11:17 a.m.Exchange Online incident EX1550543 begins in admin logs (“unable to connect to mailboxes through multiple connection methods”).Nov 12, 16:20Nov 12, 11:20 a.m.Microsoft posts initial alert inside private Microsoft 365 Admin Center consoles.Nov 12, 17:52–17:54Nov 12, 12:52–12:54 p.m.Anthropic flags elevated M365 connector errors at 17:52 UTC; two minutes later, <a href="http://twitter.com/MSFT365Status">@MSFT365Status</a> publicly acknowledges Americas Exchange issues.Nov 12, ~18:00–18:09Nov 12, ~1:00–1:09 p.m.Downdetector Outlook reports hit 128,874–138,058 across 74 countries and territories; at 18:07 UTC, Microsoft warns EX1550543 may affect “any region.”Nov 12, 18:26–18:34Nov 12, 1:26–1:34 p.m.Suite-wide incident MO1550821 begins at 18:26 UTC; at 18:34 UTC, Anthropic identifies an upstream provider issue breaking mail, calendar, and search tools.Nov 12, 19:31–19:57Nov 12, 2:31–2:57 p.m.Microsoft rules out a traffic spike, cites failed automated health checks and elevated CPU, begins server restarts, and tests a configuration workaround.Nov 12, ~20:30–23:00Nov 12, ~3:30–6:00 p.m.By 7:30 a.m. AEDT (20:30 UTC), Australia logs &gt;2,500 reports; by early U.S. evening, Downdetector Outlook reports fall roughly 75% to ~34,490.Nov 13, 01:48Nov 12, 8:48 p.m.Microsoft isolates the issue to an overwhelmed server process handling secure connections and starts rolling back correlated network/config changes.Nov 13, 03:19Nov 12, 10:19 p.m.Microsoft warns applied serviceability fixes are “not taking as expected” and seeks alternate capacity workstreams amid sustained imbalances.Nov 13, 04:23Nov 12, 11:23 p.m.Microsoft executive Charles Lamanna publicly states there is no indication of malicious or cyber activity as partial recovery continues.</p><p>(Note: OpenAI also logged an incident for its Microsoft 365 connector at 10:29 a.m. on its status page, though the captured status timestamp did not display a local timezone label.)</p><h3>The Blast Radius: How an Email Outage Broke AI Agents, Security Tools, and Public Services</h3><p>Although Exchange Online sat at the center of the storm, the outage exposed how deeply non-email systems now depend on Microsoft’s messaging and front-end infrastructure.</p><p>Impact LayerAffected Platforms and InstitutionsObserved Operational ImpactCore ProductivityExchange Online (Web, MAPI, ActiveSync, REST), Outlook, Teams, SharePoint, OneDrive, Copilot 365, Planner, To Do, Power BI</p><p>Mailbox lockouts, queued mail, broken Teams meeting scheduling and chat creation, Copilot disruptions, blocked sensitivity-labeled Power BI PDF exports.</p><p>Security and GovernanceDefender XDR, Defender for Office 365, Microsoft Intune, Microsoft Purview (DLP, Auto Labelling, Communication Compliance), Apps Admin Center</p><p>Listed under MO1550821 as potentially degraded, risking temporary visibility blind spots, delayed security alerts, and disrupted policy enforcement.</p><p>Downstream AI PlatformsOpenAI (ChatGPT Microsoft 365 connector), Anthropic (Claude MCP Microsoft 365 connector)</p><p>Broken new connector sessions and failing enterprise tools across mail, calendar, SharePoint, Teams chat, and search.</p><p>Public Sector and AcademiaUIUC, THERS (Japan), AFP, ASIO, DFAT, DAFF, Service NSW, Australian state agencies, AMA, RACGP</p><p>Broken Canvas alerts and emailed IT ticket intake; closed regional Service NSW counters; doctors reverting to paper scripts due to email-tied prescribing authentication.</p><p>The ripple into OpenAI and Anthropic underscores a new dependency across modern enterprise AI workflows. Both ChatGPT and Claude offer native connectors allowing users to search and reason directly over corporate Microsoft 365 data in Anthropic’s case, built on Model Context Protocol (MCP) endpoints authorized via delegated Microsoft Graph permissions inside Microsoft Entra ID. The moment Microsoft 365’s front-end and Exchange endpoints began timing out, both AI providers suffered immediate upstream failures across enterprise mail, calendar, and search tools.</p><p>In Australia, where the disruption arrived at the start of Friday’s workday, the public-sector fallout illustrated how single-cloud dependencies bleed into physical services. The Australian Federal Police (AFP) confirmed it was affected while domestic intelligence agency ASIO reported partial disruption, though Australian Computer Society CEO Dr. Prins Ralston noted both security agencies had fallback workarounds that small businesses and smaller agencies lacked. Federal departments including Foreign Affairs and Trade (DFAT) and Agriculture, Fisheries and Forestry (DAFF) warned clients off email, redirecting urgent sanctions inquiries to phone lines. Service NSW closed some regional customer centers and instituted priority call-backs. Most strikingly, the Australian Medical Association (AMA) and Royal Australian College of General Practitioners (RACGP) reported that while clinical care and patient safety were unharmed, doctors had to revert to paper prescriptions because electronic prescribing authentication depended on email access.</p><p>Higher education felt similar second-order breakage. At the University of Illinois Urbana-Champaign (UIUC), officials warned students and staff not only that Outlook calendars were severed from Zoom links, but that emailed IT helpdesk tickets stopped generating support cases and Canvas learning notifications disappeared.</p><h3>Was It a Cyberattack or Patch Tuesday Fallout?</h3><p>Whenever Microsoft email goes dark worldwide, speculation quickly turns to hostile interference and the November calendar offered tempting fuel. Just two days before the outage, on November 10, 2026, Microsoft shipped a massive Patch Tuesday release addressing between 1,093 and 1,252 vulnerabilities depending on how cybersecurity vendors tallied earlier-month and third-party advisories. Front and center was CVE-2026–124194, an actively exploited Outlook spoofing zero-day allowing attackers to capture NTLMv2 authentication hashes via specially crafted files (though not through the Preview Pane), which CISA swiftly added to its Known Exploited Vulnerabilities catalog with a November 24 remediation deadline for U.S. federal agencies.</p><p>Despite the timing, there is no evidence linking the November 12–13 outage to CVE-2026–124194 or any cyberattack. Microsoft’s service advisories, corporate spokespeople, executive Charles Lamanna, and Australian Home Affairs Minister Tony Burke-alongside review by Australia’s National Cyber Security Coordinator, Lt Gen Michelle McGuinness-all stated there was no indication of malicious activity.</p><p>Still, an infrastructure failure does not have to be caused by hackers to create a security hazard. The Victorian government warned residents about opportunistic scammers exploiting the outage, and cybersecurity advisories urged vigilance against phishing lures and social-engineered helpdesk requests masquerading as “Microsoft recovery steps”. Worse, because Defender XDR, Defender for Office 365, Purview, and Intune appeared on the MO1550821 affected-services list, security teams faced temporary telemetry delays and visibility gaps while managing panicked user tickets.</p><p><strong>What IT Teams and Users Should Do Right Now</strong></p><p>During a partial, wobbling recovery where restarted servers can drop back offline under live load, aggressive local troubleshooting makes recovery harder rather than faster. Microsoft support moderators and incident-response guidance draw a clear line between holding steady while EX1550543 and MO1550821 remain open and auditing thoroughly once stability returns.</p><h4><strong>While Recovery Remains Partial</strong></h4><ol><li><strong>Monitor authoritative status channels:</strong> Track EX1550543 and MO1550821 under Health &gt; Service health in the Microsoft 365 Admin Center, or use the unauthenticated status.cloud.microsoft page and @MSFT365Status on X if the admin console fails to load.</li><li><strong>Keep local Outlook configurations intact:</strong> Instruct users not to reinstall Microsoft 365 apps, delete and rebuild Outlook profiles, remove mobile accounts, or repeatedly reset passwords none of these actions resolve an upstream server connection failure.</li><li><strong>Leave security guardrails locked in place:</strong> Do not disable Multi-Factor Authentication (MFA), relax Conditional Access rules, or alter mail-flow routing as temporary workarounds, and maintain strict identity verification at the help desk despite high ticket volumes.</li><li><strong>Coordinate through out-of-band channels:</strong> Direct urgent internal communications through non-Microsoft platforms such as Zoom, Google Workspace, Slack, SMS, or direct phone lines and use web portals rather than email for critical ticketing or calendar access.</li></ol><h3>Once Full Service Returns</h3><ol><li><strong>Verify mailbox access and mail flow separately:</strong> Reopening Outlook only confirms client connectivity; do not assume every message sent during the outage arrived, as delivery depends on whether Exchange queued the email or whether sending servers timed out with Non-Delivery Reports (NDRs). Audit Exchange Mail flow &gt; Message trace and retain SMTP status codes and queue ages for unresolved traffic.</li><li><strong>Audit sign-in logs and inbox rules:</strong> Once telemetry catches up, review Entra sign-in logs for anomalous locations or unusual MFA patterns during the outage window, and check mailboxes for unauthorized forwarding rules, inbox rules, or third-party app consents.</li><li><strong>Confirm security tooling enforcement:</strong> Verify that Defender XDR, Defender for Office 365, Purview DLP policies, and Intune compliance controls are actively enforcing again and triage any delayed security alerts.</li><li><strong>Document UTC timestamps for SLA claims:</strong> Record exact first-observed failures, affected connection methods, and final recovery times in UTC to support any service-credit requests under your Microsoft Service Level Agreement.</li></ol><h3>Hidden Dependencies in a Single-Cloud World</h3><p>Step back from the 138,000-plus Downdetector reports, and the November 2026 outage fits a troubling pattern. Over the past sixteen months alone, enterprise users have weathered a 19-to-21-hour global Outlook outage in July 2025, a major Azure disruption in October 2025, a multi-hour Microsoft 365 outage in January 2026, and an August 2026 code-rollout breakdown that required 67 hours nearly three days to fully resolve.</p><p>What makes November 12–13 especially revealing is how thoroughly both legacy workflows and brand-new AI stacks inherit Exchange Online’s availability assumptions. When a secure-connection process on a cloud email tier buckles, it no longer merely delays internal memos. It forces clinics onto paper prescriptions, halts university helpdesk automation, obscures visibility in XDR consoles, and severs the connectors powering ChatGPT and Claude across enterprise knowledge bases. Until organizations deliberately decouple emergency communications, authentication workflows, security telemetry, and AI tooling from single-vendor email infrastructure, every automated health-check cascade inside Microsoft’s data centers will continue to ripple straight into the daily mechanics of the global economy.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=402105a8cc46" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[PHP2Shell | How an AI Uncovered a 10-Year-Old Flaw Threatening 70% of the Web]]></title>
            <link>https://medium.com/@thinuridulsini/php2shell-how-an-ai-uncovered-a-10-year-old-flaw-threatening-70-of-the-web-959aba60a1f7?source=rss-2e65eb35360b------2</link>
            <guid isPermaLink="false">https://medium.com/p/959aba60a1f7</guid>
            <category><![CDATA[remote-code-execution]]></category>
            <category><![CDATA[php2shell]]></category>
            <category><![CDATA[cve-2026-109618]]></category>
            <category><![CDATA[php]]></category>
            <dc:creator><![CDATA[Thinuri Wickramarachchi]]></dc:creator>
            <pubDate>Fri, 23 Oct 2026 02:48:08 GMT</pubDate>
            <atom:updated>2026-10-23T02:48:08.795Z</atom:updated>
            <content:encoded><![CDATA[<blockquote><em>CVE-2026–109618 &amp; CVE-2026–109619: how memory limits, longjmp(), and stale states allowed unauthenticated Remote Code Execution.</em></blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*lP-8z3MGdyKHaCA6DpnPxw.png" /></figure><p>If you bought something online today, read a blog post, or logged into a customer portal, there is roughly a 70% chance that the code serving your request was written in <strong>PHP</strong>.</p><p>For decades, PHP has been the engine behind major Content Management Systems (CMS) like WordPress, Drupal, and Joomla. But in October 2026, cybersecurity researchers unveiled <strong>PHP2Shell</strong> a pair of critical, zero-day vulnerabilities (<strong>CVE-2026–109618</strong> and <strong>CVE-2026–109619</strong>) with a CVSS score of 9.2.</p><p>What makes PHP2Shell terrifying isn’t just that it allows an unauthenticated attacker to execute code on a remote web server. It is <em>how</em> it happens:</p><ol><li><strong>No Authentication Required:</strong> An attacker doesn’t need a username or password.</li><li><strong>Invisible in Logs:</strong> Standard web application logs show a normal <strong>HTTP 200 OK</strong>.</li><li><strong>Pre-Execution Exploitation:</strong> The attack can trigger in memory <em>before</em> your website’s security plugins or code ever run.</li><li><strong>AI-Discovered:</strong> It was uncovered using AI-assisted source code auditing, finding memory flaws that sat hidden in PHP core for over a decade.</li></ol><p>To understand how this works, we need to go back to the basics of how PHP handles memory, errors, and requests.</p><h3>1. The Setting: How Web Servers Think</h3><p>Imagine a web server as a restaurant kitchen, and PHP-FPM (FastCGI Process Manager) as a team of cooks (worker processes).</p><p>When a visitor loads a page on your website, a request arrives at the kitchen.</p><ol><li><strong>The Request Arrives:</strong> A worker process takes the request (e.g., POST form data or file upload).</li><li><strong>Memory Allocation:</strong> The cook reserves a cutting board (allocates temporary memory) to parse the incoming data.</li><li><strong>Execution:</strong> PHP runs your scripts (WordPress plugins, database calls).</li><li><strong>Cleanup:</strong> Once the page is rendered and sent to the visitor, the cook wipes the cutting board clean, resets everything, and waits for the <em>next</em> customer request.</li></ol><p>To prevent a single rogue request from consuming the entire server’s RAM, PHP enforces a safety guardrail called memory_limit (e.g., 128M). If a single request tries to use more RAM than allowed, PHP panics, throws an Out-Of-Memory (OOM) error, and terminates the request.</p><h3>2. The Trap: The longjmp() Escape Hatch</h3><p>How does PHP handle a sudden Out-Of-Memory error?</p><p>In lower-level code like C (which PHP itself is written in), when a fatal error or memory shortage occurs, the program cannot just continue. It needs to jump back to a safe recovery point. In C programming, this emergency jump is done using a feature called setjmp() and longjmp().</p><p>Think of longjmp() like a <strong>teleporter button</strong>.</p><ul><li>When PHP starts processing a request, it places a “teleport pad” (setjmp).</li><li>If at any point during request parsing the server runs out of memory, PHP presses the button (longjmp) and instantly warps execution back to the error handler to clean up and abort.</li></ul><p>Here lies the fatal flaw.</p><h3>Vulnerability #1: CVE-2026–109618 (Uninitialized Cleanup)</h3><p>When a complex HTTP request comes in (like a large file upload or deeply nested POST form data), PHP initializes internal data structures step by step:</p><h4><em>Step 1: Mark structure as “Ready” →Step 2: Allocate Memory →Step 3: Fill with Valid Pointers</em></h4><p>If an attacker deliberately crafts a request that forces PHP to hit its memory_limit right between <strong>Step 1</strong> and <strong>Step 3</strong>, PHP triggers a longjmp().</p><p>PHP warps instantly to the cleanup phase. The cleanup routine looks at the state, sees <em>“Oh, Step 1 said this data structure is Ready!”</em>, and attempts to free memory referenced by pointers in that structure. But because Step 3 never happened, those pointers contain <strong>garbage memory addresses</strong> or <strong>null values</strong>.</p><p>Trying to read or free garbage memory addresses leads to <strong>memory corruption</strong>, giving an attacker the ability to manipulate execution flow and gain remote code execution (RCE).</p><h3>3. The Zombie Worker: Stale State Across Requests</h3><h3>Vulnerability #2: CVE-2026–109619 (Stale Global State)</h3><p>Web servers recycle worker processes to save computer memory. A single PHP worker process might handle 1,000 requests before restarting.</p><p>When an Out-Of-Memory error triggers a longjmp(), PHP aborts the current request. However, because the normal execution flow was abruptly skipped, <strong>global variables and pointers pointing to freed memory survive</strong>.</p><p>When the <em>next</em> innocent customer (or the attacker’s second request) arrives at that same worker process:</p><ol><li>The worker process accepts the new request.</li><li>But its internal memory management still holds pointers pointing to memory allocated during the <em>previous</em> failed request (which has now been freed).</li><li>This creates a <strong>Use-After-Free (UAF)</strong> scenario.</li></ol><p>An attacker can deliberately send Request A to corrupt and poison the worker’s internal state, and then immediately send Request B to hijack the stale state, overwrite function pointers, and execute arbitrary code on the operating system.</p><h3>4. The Attack Sequence: Step-by-Step</h3><p>Let’s look at how an attack unfolds in practice:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*L84gTWSgCgsMNy9TaB-JpA.png" /></figure><h3>Why standard security fails to detect it:</h3><ul><li><strong>No Application Logs:</strong> PHP crashes internally during the engine lifecycle or handles the error via fallback paths before your application (like WordPress) can log an exception.</li><li><strong>Returns HTTP 200:</strong> Depending on how web servers like Apache or Nginx communicate with the PHP worker process during certain memory error states, the front-facing web server may log an HTTP 200 OK response while the background worker process is exploited.</li></ul><h3>5. How It Was Revealed: The AI Auditing Breakthrough</h3><p>For over 10 years, these memory safety patterns went unnoticed by human auditors because they sit at the intersection of complex C macros, C library error-handling routines (setjmp/longjmp), and PHP’s garbage collection.</p><p>In mid-2026, researchers leveraged advanced AI models trained on system-level code auditing to analyze the core PHP C source code. The AI was instructed to map every execution path where longjmp() could interrupt memory state initialization.</p><p>The AI identified that PHP’s request-lifecycle memory guarantees did not hold true during sudden OOM interruptions, leading directly to the discovery of PHP2Shell.</p><h3>6. How to Protect Your Applications</h3><p>Because these flaws are inside the <strong>PHP core interpreter</strong>, application-level fixes (like updating WordPress plugins) <strong>will not protect you</strong>. You must fix the underlying runtime engine.</p><h4>1. Upgrade PHP Immediately</h4><p>The official PHP security team released emergency patches. Update your PHP installation to one of the following safe versions:</p><ul><li><strong>PHP 8.5.12</strong> or higher</li><li><strong>PHP 8.4.27</strong> or higher</li><li><strong>PHP 8.3.36</strong> or higher</li><li><strong>PHP 8.2.35</strong> or higher</li></ul><p><em>(Note: PHP 7.x and 8.1 are End-of-Life (EOL) and do not receive official full upstream releases; legacy systems require OS vendor backports or migrating to supported PHP versions).</em></p><h4>2. Mitigation Strategies (If patching is delayed)</h4><p>If you cannot upgrade immediately, apply these server configuration changes to reduce memory pressure and limit attack vectors:</p><ul><li><strong>Remove finite PHP memory limits for workers (Use with caution):</strong> Setting memory_limit = -1 in php.ini prevents PHP&#39;s internal OOM handler from triggering. <em>Warning:</em> You must enforce memory bounds at the OS level (e.g., Linux cgroup v2 or systemd limits) so your server doesn&#39;t run out of system RAM.</li><li><strong>Reduce Input Pressure:</strong></li><li>; Lower maximum POST request size post_max_size = 8M ; Disable session upload progress if unused session.upload_progress.enabled = 0 ; Turn off public error display display_errors = Off</li><li><strong>Recycle Workers Frequently:</strong> Lower pm.max_requests in your PHP-FPM configuration (e.g., pm.max_requests = 500). This forces worker processes to destroy themselves and restart more often, wiping any stale global state before it can be exploited.</li></ul><h3>Key Takeaway</h3><p>PHP2Shell serves as a stark reminder of modern software complexity. Even decades-old, heavily tested open-source engines can harbor memory-safety flaws deep in their low-level runtime lifecycle. As AI tools lower the barrier to discovering these intricate edge cases, maintaining rapid patching and defense-in-depth monitoring (like EDR and OS-level containment) is more critical than ever.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=959aba60a1f7" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Old Money | More Than a Fashion Trend — The Fascinating History, Hidden Truths, and Timeless Appeal…]]></title>
            <link>https://medium.com/@thinuridulsini/old-money-more-than-a-fashion-trend-the-fascinating-history-hidden-truths-and-timeless-appeal-49891ff8da98?source=rss-2e65eb35360b------2</link>
            <guid isPermaLink="false">https://medium.com/p/49891ff8da98</guid>
            <dc:creator><![CDATA[Thinuri Wickramarachchi]]></dc:creator>
            <pubDate>Thu, 02 Jul 2026 03:00:57 GMT</pubDate>
            <atom:updated>2026-07-02T03:00:57.463Z</atom:updated>
            <content:encoded><![CDATA[<h3>Old Money | More Than a Fashion Trend — The Fascinating History, Hidden Truths, and Timeless Appeal Behind the World’s Most Misunderstood Aesthetic</h3><blockquote><em>“It’s not about wearing a beige sweater or driving a vintage car. It’s about a story that began centuries before social media ever existed.”</em></blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*wT5f3yN7bEkGrUGHF5zPRA.png" /></figure><p>Every few months, a new fashion trend takes over the internet.</p><p>One year it’s “clean girl.” The next it’s “quiet luxury.” Then suddenly, everyone is talking about <strong>Old Money</strong>.</p><p>Pinterest boards fill with grand libraries, tailored blazers, Oxford shoes, linen shirts, country estates, vintage watches, and tennis courts bathed in afternoon sunlight. TikTok creators promise to teach you how to “dress old money,” while luxury brands quietly embrace the trend.</p><p>But here’s the question most people never ask:</p><p><strong>What exactly is Old Money?</strong></p><p>Is it simply expensive clothing?</p><p>A fashion aesthetic?</p><p>A lifestyle?</p><p>Or is it something much deeper something rooted in centuries of history, power, privilege, and even controversy?</p><p>The answer is far more fascinating than a wardrobe.</p><h3>Long Before Instagram…</h3><p>The story of Old Money didn’t begin on social media.</p><p>It didn’t even begin in America.</p><p>It began hundreds of years ago.</p><p>Imagine Europe during the Middle Ages.</p><p>Kings ruled kingdoms.</p><p>Nobles owned vast lands.</p><p>Castles stood on hills overlooking villages that had existed for generations.</p><p>Families didn’t become wealthy because they launched startups.</p><p>They inherited estates, farms, businesses, trading rights, and political influence from their ancestors.</p><p>In England, families passed down titles like Duke, Earl, or Baron.</p><p>In Italy, merchant families such as the <strong>Medici</strong> transformed banking into immense wealth and political influence during the Renaissance.</p><p>Across Europe, fortunes weren’t measured by annual income.</p><p>They were measured by <strong>how many generations had protected and expanded them.</strong></p><p>The wealth itself became part of the family’s identity.</p><p>This was the earliest form of what we now describe as <strong>Old Money</strong>.</p><h3>Then Came America</h3><p>The phrase <strong>“Old Money”</strong> became widely used in the United States during the late nineteenth century.</p><p>America was changing rapidly.</p><p>Factories appeared everywhere.</p><p>Railroads connected entire states.</p><p>Oil, steel, banking, and industry created fortunes that the world had never seen before.</p><p>Entrepreneurs like John D. Rockefeller, Andrew Carnegie, Cornelius Vanderbilt, and J.P. Morgan became extraordinarily wealthy within a single lifetime.</p><p>They represented something entirely new.</p><p>They were called <strong>New Money.</strong></p><p>At the same time, America already had wealthy families whose fortunes stretched back generations.</p><p>These established families viewed themselves differently.</p><p>To them, wealth wasn’t just money.</p><p>It was heritage.</p><p>Reputation.</p><p>Education.</p><p>Tradition.</p><p>Social standing.</p><p>Thus, society began distinguishing between <strong>Old Money</strong> and <strong>New Money</strong>.</p><h3>The Difference Isn’t the Amount of Money</h3><p>Here’s something many people misunderstand.</p><p>Old Money doesn’t necessarily mean richer.</p><p>A young entrepreneur worth several billion dollars may have far more money than an old aristocratic family.</p><p>Yet society still labels one as New Money and the other as Old Money.</p><p>Why?</p><p>Because the distinction isn’t about <strong>how much</strong> wealth exists.</p><p>It’s about <strong>where it came from and how long it has survived.</strong></p><p>If wealth survives multiple generations while preserving a family’s identity and legacy, it gradually becomes Old Money.</p><p>Time itself becomes part of the fortune.</p><h3>Wealth That Survives Is Surprisingly Rare</h3><p>Creating wealth is difficult.</p><p>Keeping it for centuries is even harder.</p><p>Financial historians often reference a pattern found in many families around the world:</p><ul><li>The first generation creates wealth.</li><li>The second generation expands it.</li><li>The third generation often struggles to preserve it.</li></ul><p>Although this isn’t a universal law, many fortunes decline because of poor financial decisions, inheritance disputes, changing economies, wars, or family conflict.</p><p>Families that remain wealthy across four, five, or even ten generations are relatively uncommon.</p><p>This rarity contributes to the mystique surrounding Old Money.</p><h3>Why Does Old Money Feel So “Classy”?</h3><p>If you’ve ever looked at Old Money fashion online, you may have noticed something interesting.</p><p>It rarely screams for attention.</p><p>Instead of giant luxury logos, you see:</p><ul><li>Tailored navy blazers</li><li>Linen shirts</li><li>Cashmere sweaters</li><li>Leather loafers</li><li>Pearl earrings</li><li>Vintage watches</li><li>Neutral colors</li><li>Simple elegance</li></ul><p>This reflects an idea sometimes called <strong>“quiet luxury.”</strong></p><p>Historically, established wealthy families often didn’t need visible symbols to prove they belonged to the upper class.</p><p>Their reputation already spoke for itself.</p><p>This doesn’t mean every wealthy family dressed this way, nor that understated fashion is exclusive to inherited wealth. But over time, these visual cues became associated with long-established privilege.</p><h3>It’s More Than Clothing</h3><p>The Old Money image also includes certain habits and values that have been romanticized over time.</p><p>People often associate it with:</p><ul><li>Reading and lifelong learning</li><li>Respect for tradition</li><li>Good manners</li><li>Long-term investing</li><li>Philanthropy</li><li>Classical architecture</li><li>Appreciation for art and history</li><li>Financial discipline</li><li>Privacy</li></ul><p>Whether every Old Money family embodies these traits is another question but these qualities have become part of the cultural image.</p><h3>Then Social Media Changed Everything</h3><p>Around the early 2020s, something unexpected happened.</p><p>TikTok.</p><p>Instagram.</p><p>Pinterest.</p><p>Suddenly, millions of people discovered an aesthetic they had never heard of before.</p><p>Creators started posting videos titled:</p><ul><li>“How to Dress Old Money”</li><li>“Old Money Morning Routine”</li><li>“Old Money Hairstyles”</li><li>“Old Money Rules”</li></ul><p>The internet transformed a centuries-old social concept into a fashion trend.</p><p>Ironically, many influencers promoting the style had no connection to inherited wealth at all.</p><p>And that’s perfectly fine.</p><p>Fashion evolves by borrowing inspiration.</p><p>Today, “Old Money” often describes a visual style rather than a family’s history.</p><h3>Why Are People Drawn to It?</h3><p>Perhaps because it represents something people feel is missing.</p><p>Modern life moves quickly.</p><p>Social media rewards instant gratification.</p><p>Algorithms encourage constant attention.</p><p>The Old Money aesthetic offers the opposite image.</p><p>It suggests:</p><p>Slow mornings.</p><p>Handwritten letters.</p><p>Historic libraries.</p><p>Timeless clothing.</p><p>Elegant homes.</p><p>Long-term thinking.</p><p>Whether this image reflects reality is another matter, but emotionally it appeals to many people.</p><p>It feels calm in a noisy world.</p><h3>The Hidden Side of the Story</h3><p>This is where the conversation becomes more complex.</p><p>Not every historic fortune has a dark origin.</p><p>But history also tells us that some did.</p><p>Across different periods, certain fortunes grew through systems we now recognize as deeply unjust.</p><p>Examples include:</p><ul><li>Colonial expansion</li><li>The transatlantic slave trade</li><li>Exploitative labor systems</li><li>Land confiscation</li><li>Political favoritism</li><li>Profits from war</li></ul><p>Many banks, trading companies, plantations, and commercial enterprises accumulated wealth within these historical systems.</p><p>Modern historians continue researching how these systems contributed to the fortunes of specific families and institutions.</p><p>This doesn’t mean every Old Money family shares the same history.</p><p>Each family’s story is different.</p><p>Some fortunes grew through banking, engineering, manufacturing, agriculture, entrepreneurship, scientific innovation, or careful investing.</p><p>History is rarely simple.</p><h3>Why Do People Say “Every Old Fortune Has Blood on It”?</h3><p>You’ve probably seen comments online claiming:</p><blockquote><em>“There’s no such thing as clean Old Money.”</em></blockquote><p>This statement is emotionally powerful, but historically too broad.</p><p>A more accurate conclusion is:</p><p>Some fortunes were connected to exploitation.</p><p>Some were not.</p><p>Many existed somewhere between these extremes, shaped by the legal, political, and economic systems of their time.</p><p>History requires evidence, not assumptions.</p><p>Each family’s past deserves to be examined on its own merits.</p><h3>The Power of Family Legacy</h3><p>One fascinating aspect of Old Money has little to do with money itself.</p><p>It’s legacy.</p><p>Imagine two families.</p><p>The first earns millions in one generation but loses everything within twenty years.</p><p>The second experiences financial setbacks, rebuilds, educates each generation about stewardship, and preserves its identity over centuries.</p><p>The second family’s story often leaves a stronger cultural impression.</p><p>Old Money isn’t only about inheritance.</p><p>It’s also about continuity.</p><h3>Can Anyone Become Old Money?</h3><p>Not today.</p><p>But perhaps one day.</p><p>Every Old Money family started with someone.</p><p>Someone opened a business.</p><p>Someone invested wisely.</p><p>Someone took risks.</p><p>Someone built the first fortune.</p><p>If later generations preserve that wealth responsibly while maintaining a lasting family legacy, future historians may describe them as Old Money.</p><p>The label is earned through time not purchased overnight.</p><h3>What Can We Learn From It?</h3><p>The internet often reduces Old Money to fashion.</p><p>History shows it’s much more.</p><p>It is a story about economics.</p><p>Family.</p><p>Culture.</p><p>Power.</p><p>Tradition.</p><p>Opportunity.</p><p>Responsibility.</p><p>And sometimes, injustice.</p><p>The greatest lesson isn’t to imitate someone’s wardrobe.</p><p>It’s to understand how wealth is created, preserved, and remembered.</p><p>Because true legacy isn’t measured by the price of a watch or the logo on a sweater.</p><p>It’s measured by what survives when fashions change, markets crash, and generations pass.</p><p>Perhaps that’s why the idea of Old Money continues to fascinate us.</p><p>Not because of what people wear.</p><p>But because it asks a timeless question:</p><p><strong>If your family were remembered 200 years from now, what legacy would you leave behind?</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=49891ff8da98" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Is the Universe a Hologram? The Fascinating Story Behind One of Physics’ Most Mind-Bending Ideas]]></title>
            <link>https://medium.com/@thinuridulsini/is-the-universe-a-hologram-the-fascinating-story-behind-one-of-physics-most-mind-bending-ideas-5409e27d970d?source=rss-2e65eb35360b------2</link>
            <guid isPermaLink="false">https://medium.com/p/5409e27d970d</guid>
            <dc:creator><![CDATA[Thinuri Wickramarachchi]]></dc:creator>
            <pubDate>Thu, 25 Jun 2026 03:31:01 GMT</pubDate>
            <atom:updated>2026-06-25T03:31:01.167Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*FqiXAm0wP3yYBEyV4jQH8w.png" /></figure><h3>Imagine This…</h3><p>You are watching a movie projected onto a screen.</p><p>The characters see a three-dimensional world. They experience depth, distance, movement, and time. To them, everything appears real and complete.</p><p>But from your perspective, everything they experience comes from information encoded on a flat two-dimensional surface.</p><p>Now imagine something even stranger:</p><p>What if our entire universe worked in a similar way?</p><p>What if the three-dimensional world we experience every day every star, galaxy, planet, and even ourselves is actually the projection of information stored on a distant cosmic boundary?</p><p>This is the astonishing idea behind the <strong>Holographic Universe Theory</strong>, one of the most intriguing concepts in modern theoretical physics.</p><p>It sounds like science fiction, but it emerged from some of the most serious scientific attempts to understand black holes, gravity, quantum mechanics, and the fundamental nature of reality.</p><p>Let’s explore how this idea appeared, why scientists take it seriously, what evidence supports it, and what it could mean for humanity.</p><h3>The Mystery That Started Everything</h3><p>For centuries, scientists believed that information about objects was contained within their volume.</p><p>A bigger box could store more things.</p><p>A larger room could hold more people.</p><p>A larger hard drive could store more data.</p><p>This seems obvious.</p><p>Then black holes entered the story.</p><p>Black holes are regions of space where gravity becomes so strong that nothing not even light can escape.</p><p>For a long time, physicists thought black holes simply swallowed everything that fell into them.</p><p>But this created a huge problem.</p><p>According to quantum mechanics, information can never truly disappear.</p><p>So what happens to the information contained in objects that fall into a black hole?</p><p>This question became one of the greatest mysteries in physics.</p><h3>Stephen Hawking Changes Everything</h3><p>In the 1970s, physicist Stephen Hawking made a shocking discovery.</p><p>He showed that black holes are not completely black.</p><p>They emit a tiny amount of radiation, now known as <strong>Hawking Radiation</strong>.</p><p>This meant black holes could slowly evaporate over incredibly long periods of time.</p><p>But a new puzzle emerged.</p><p>If a black hole eventually disappears, what happens to all the information that fell inside it?</p><p>If the information vanishes, quantum mechanics breaks.</p><p>If it survives, where is it stored?</p><p>Scientists called this the <strong>Black Hole Information Paradox</strong>.</p><p>Solving this paradox would eventually lead to the holographic principle.</p><h3>The Strange Discovery About Information</h3><p>Physicist Jacob Bekenstein made an important observation.</p><p>He discovered that the amount of information a black hole can contain is proportional not to its volume but to the surface area of its event horizon.</p><p>This was shocking.</p><p>Imagine buying a hard drive whose storage capacity depended on the size of its outer casing rather than the space inside.</p><p>That seems impossible.</p><p>Yet mathematics suggested exactly that.</p><p>The entropy or information content of a black hole is determined by its surface area.</p><p>This seemingly small discovery would change theoretical physics forever.</p><h3>Enter the Holographic Principle</h3><p>In the 1990s, physicists Gerard ‘t Hooft and Leonard Susskind proposed a radical idea.</p><p>What if this wasn’t just true for black holes?</p><p>What if it was true for the entire universe?</p><p>What if all information contained within a volume of space is actually encoded on its boundary surface?</p><blockquote>This became known as the Holographic Principle.</blockquote><p>According to this principle:</p><blockquote><em>Everything happening inside a region of space can be described by information stored on the boundary surrounding that region.</em></blockquote><p>In simple terms, our three-dimensional reality may emerge from information existing on a two-dimensional surface.</p><p>Just like a <em>hologram</em>.</p><h3>But What Is a Hologram?</h3><p>A hologram is a flat surface that stores information about a three-dimensional image.</p><p>When illuminated correctly, it produces the illusion of depth.</p><p>Although the image appears three-dimensional, the information is encoded in two dimensions.</p><p>The holographic universe idea uses this concept as an analogy.</p><p>It does not mean the universe is fake.</p><p>It means that the information describing our universe might fundamentally exist in fewer dimensions than we perceive.</p><h3>The Mathematical Breakthrough</h3><p>The holographic principle remained an intriguing idea until 1997.</p><p>Then physicist Juan Maldacena introduced one of the most important discoveries in modern theoretical physics.</p><p>It became known as:</p><h4>AdS/CFT Correspondence</h4><p>The name sounds intimidating, but the core idea is surprisingly elegant.</p><p>Maldacena showed that:</p><p>A gravity theory existing in a higher-dimensional space can be mathematically equivalent to a quantum theory existing on a lower-dimensional boundary.</p><p>Two completely different descriptions produced identical results.</p><p>It was as if two different languages were describing the same reality.</p><p>This provided the first concrete mathematical evidence that holographic descriptions could actually work.</p><p>Many physicists consider it one of the greatest breakthroughs since Einstein’s theory of relativity.</p><h3>The Science Behind It</h3><p>To understand why the holographic principle is important, we need to examine the biggest problem in modern physics.</p><p>Today, science relies on two highly successful theories.</p><h3>General Relativity</h3><p>Developed by Albert Einstein.</p><p>It explains:</p><ul><li>Gravity</li><li>Planets</li><li>Stars</li><li>Black holes</li><li>The structure of the universe</li></ul><p>It works exceptionally well at large scales.</p><h3>Quantum Mechanics</h3><p>It explains:</p><ul><li>Atoms</li><li>Electrons</li><li>Subatomic particles</li><li>Quantum fields</li></ul><p>It works exceptionally well at tiny scales.</p><h3>The Problem</h3><p>When scientists try to combine quantum mechanics and gravity, the mathematics often breaks down.</p><p>The universe appears to speak two different languages.</p><p>Finding a theory that unifies both is one of the greatest goals in physics.</p><p>The holographic principle offers a possible bridge.</p><p>By translating gravitational systems into quantum systems, it may provide clues toward a long-sought <strong><em>“Theory of Everything.”</em></strong></p><h3>Is There Any Evidence?</h3><p>This is where things become complicated.</p><p>The holographic principle has strong mathematical support.</p><p>Many calculations work beautifully.</p><p>Researchers use holographic methods to study:</p><ul><li>Black holes</li><li>Quantum information</li><li>Particle physics</li><li>Condensed matter systems</li><li>Quantum gravity</li></ul><p>However, direct experimental proof remains elusive.</p><p>No experiment has yet confirmed that our universe is literally holographic.</p><p>Scientists continue searching for observable signatures.</p><p>So the idea remains a powerful hypothesis rather than an established fact.</p><h3>Could Reality Be Information?</h3><p>One of the most profound implications of holographic thinking is that information may be more fundamental than matter.</p><p>Traditionally we think:</p><p>Matter creates information.</p><p>But holographic theories suggest:</p><p>Information may create matter.</p><p>This shifts our understanding of reality.</p><p>Space, time, gravity, and particles might emerge from deeper informational structures.</p><p>Some physicists describe spacetime itself as an emergent phenomenon rather than a fundamental one.</p><h3>How Does This Relate to Society?</h3><p>The holographic principle extends beyond physics.</p><p>It has influenced philosophy, technology, and popular culture.</p><p>Many people hear about the holographic universe and immediately think of:</p><ul><li>Virtual reality</li><li>Simulations</li><li>The Matrix</li><li>Digital worlds</li></ul><p>Although these ideas are not the same, they share a common theme:</p><p>The reality we experience may not be the deepest layer of existence.</p><p>This challenges how humans think about:</p><ul><li>Consciousness</li><li>Perception</li><li>Reality</li><li>Knowledge</li><li>Identity</li></ul><p>The theory encourages intellectual humility.</p><p>It reminds us that what appears obvious may not reflect the universe’s deepest truths.</p><h3>Common Misconceptions</h3><h3>Myth 1: The Universe Is Fake</h3><p>Not true.</p><p>The holographic principle does not claim reality is an illusion.</p><p>Your experiences remain physically real.</p><p>The theory concerns how information is encoded.</p><h3>Myth 2: Scientists Have Proven We Live in a Hologram</h3><p>Not true.</p><p>The idea is scientifically respected but not experimentally confirmed.</p><h3>Myth 3: It Is the Same as Simulation Theory</h3><p>Not exactly.</p><p>Simulation theory suggests reality is running on a computer.</p><p>The holographic principle is a mathematical and physical framework describing how information and spacetime might be related.</p><p>They are different concepts.</p><h3>Why Scientists Are Excited</h3><p>The holographic principle has survived decades of scrutiny.</p><p>It has produced useful mathematical tools.</p><p>It has deepened our understanding of black holes.</p><p>It has generated new approaches to quantum gravity.</p><p>Most importantly, it hints at a profound possibility:</p><p>The universe may be far stranger than our everyday intuition suggests.</p><p>The world we experience could be an emergent projection of deeper informational structures that we are only beginning to understand.</p><h3>Final Thoughts</h3><p>Few scientific ideas are as breathtaking as the holographic universe.</p><p>It emerged from black holes.</p><p>It challenged our assumptions about space and information.</p><p>It inspired new mathematical discoveries.</p><p>And it continues to shape the search for a unified understanding of reality.</p><p>Whether the universe ultimately turns out to be holographic or not, the journey itself reveals something remarkable about science.</p><p>The greatest discoveries often begin with a simple question:</p><blockquote><em>“What if reality is not what it seems?”</em></blockquote><p>The holographic principle is humanity’s attempt to answer that question.</p><p>And so far, the answer is more fascinating than anyone imagined.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=5409e27d970d" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Weak Force | The Strange Power That Changes Matter Itself]]></title>
            <link>https://medium.com/@thinuridulsini/the-weak-force-the-strange-power-that-changes-matter-itself-396e3e4f6d40?source=rss-2e65eb35360b------2</link>
            <guid isPermaLink="false">https://medium.com/p/396e3e4f6d40</guid>
            <dc:creator><![CDATA[Thinuri Wickramarachchi]]></dc:creator>
            <pubDate>Wed, 24 Jun 2026 02:00:04 GMT</pubDate>
            <atom:updated>2026-06-24T02:00:04.163Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*01Ryq6olKVPNpic1ydjx5g.png" /></figure><p>Imagine a universe where nothing could change.</p><p>The Sun would stop shining.</p><p>Radioactive elements would remain frozen forever.</p><p>Many of the atoms that make up your body would never have formed.</p><p>Such a universe would exist if one of nature’s four fundamental forces disappeared the <strong>weak nuclear force</strong>, usually called the <strong>weak interaction</strong>.</p><p>Most people have heard of gravity. Many know about electromagnetism. Some have even heard of the strong nuclear force that holds atomic nuclei together.</p><p>But the weak force is different.</p><p>It doesn’t merely pull or push.</p><p>It <strong>changes one particle into another</strong>.</p><p>And that makes it one of the most fascinating forces in physics.</p><h3>The Four Forces That Run the Universe</h3><p>As far as we know, everything in nature is controlled by four fundamental forces:</p><ol><li>Gravity</li><li>Electromagnetism</li><li>Strong Nuclear Force</li><li>Weak Nuclear Force</li></ol><p>Each force has its own job.</p><p>Gravity keeps planets in orbit.</p><p>Electromagnetism controls electricity, magnetism, chemistry, and light.</p><p>The strong force glues atomic nuclei together.</p><p>The weak force does something unique:</p><p><strong>It allows particles to transform into different particles.</strong></p><h3>The First Clue: A Strange Kind of Radioactivity</h3><p>The story begins in the late 1800s.</p><p>Scientists discovered that some atoms spontaneously emitted particles and changed into other elements.</p><p>This process became known as <strong>radioactive decay</strong>.</p><p>One particularly puzzling form was called <strong>beta decay</strong>.</p><p>During beta decay, a neutron inside an atom suddenly transformed into a proton and emitted an electron.</p><p>Scientists observed:</p><p><strong>Neutron → Proton + Electron</strong></p><p>But something didn’t add up.</p><p>The energy calculations seemed wrong.</p><p>It looked as if energy was disappearing.</p><p>And according to physics, energy cannot simply vanish.</p><p>This mystery puzzled physicists for decades.</p><h3>The Missing Particle Mystery</h3><p>In 1930, Austrian physicist Wolfgang Pauli proposed a daring idea.</p><p>Maybe there was another particle escaping unnoticed.</p><p>A particle with almost no mass.</p><p>A particle with no electric charge.</p><p>A particle that barely interacts with matter.</p><p>He called it the <strong>neutrino</strong>.</p><p>Many physicists thought the idea was crazy because nobody had ever seen such a particle.</p><p>Yet Pauli’s explanation solved the missing-energy problem.</p><p>Today we know the full beta decay process is:</p><p><strong>Neutron → Proton + Electron + Antineutrino</strong></p><p>The missing particle was real.</p><p>And the force responsible for this transformation was the weak interaction.</p><h3>What Is Actually Happening Inside a Neutron?</h3><p>To understand the weak force, we must look deeper.</p><p>Neutrons are not fundamental particles.</p><p>They are made of smaller particles called <strong>quarks</strong>.</p><p>A neutron contains:</p><ul><li>1 up quark</li><li>2 down quarks</li></ul><p>Physicists write this as:</p><p><strong>udd</strong></p><p>A proton contains:</p><ul><li>2 up quarks</li><li>1 down quark</li></ul><p>Written as:</p><p><strong>uud</strong></p><p>Notice something interesting.</p><p>Only one quark needs to change.</p><p>A down quark must become an up quark.</p><p>The weak force is the mechanism that allows this transformation.</p><h3>The Weak Force Changes Particle Identity</h3><p>Imagine particles carrying invisible identity cards.</p><p>Most forces leave those identities unchanged.</p><p>Gravity doesn’t turn an electron into something else.</p><p>Electromagnetism doesn’t change particle types.</p><p>But the weak force can.</p><p>A down quark can become an up quark.</p><p>An electron can become a neutrino.</p><p>One particle species can transform into another.</p><p>Physicists call this changing <strong>flavor</strong>.</p><p>The weak force is therefore known as the force that changes particle flavor.</p><p>This is its defining characteristic.</p><h3>How Does the Transformation Happen?</h3><p>Every force has messenger particles.</p><p>For example:</p><ul><li>Electromagnetism uses photons.</li><li>The strong force uses gluons.</li></ul><p>The weak force uses three messenger particles:</p><ul><li>W⁺ boson</li><li>W⁻ boson</li><li>Z⁰ boson</li></ul><p>These are called <strong>gauge bosons</strong>.</p><p>During beta decay, a down quark emits a W⁻ boson.</p><p>The down quark becomes an up quark.</p><p>The W⁻ boson quickly decays into:</p><ul><li>an electron</li><li>an antineutrino</li></ul><p>The entire process happens in a tiny fraction of a second.</p><p>Yet it changes a neutron into a proton.</p><h3>Why Is It Called the Weak Force?</h3><p>The answer lies in its strength.</p><p>Compared with the strong nuclear force, the weak force is much less powerful.</p><p>But the name is slightly misleading.</p><p>The weak force is still enormously stronger than gravity at the particle level.</p><p>The real reason it seems weak is because its messenger particles are extremely heavy.</p><p>For comparison:</p><ul><li>Photon mass = 0</li><li>W boson mass ≈ 80 times the proton mass</li><li>Z boson mass ≈ 91 times the proton mass</li></ul><p>Because these messenger particles are so massive, they can travel only extremely short distances.</p><p>The weak force acts across distances much smaller than an atomic nucleus.</p><p>Beyond that range, it effectively disappears.</p><h3>The Weak Force Powers the Sun</h3><p>Perhaps the most astonishing role of the weak interaction occurs inside stars.</p><p>Deep inside the Sun, hydrogen nuclei collide at incredible speeds.</p><p>For fusion to begin, one proton must eventually transform into a neutron.</p><p>That transformation is only possible because of the weak force.</p><p>Without it:</p><ul><li>Hydrogen could not fuse efficiently.</li><li>The Sun would not generate energy.</li><li>Sunlight would not exist.</li><li>Life on Earth would never have evolved.</li></ul><p>Every sunrise is partly a consequence of the weak interaction.</p><h3>The Discovery of the Weak Force</h3><p>The understanding of the weak force developed gradually.</p><h3>1896 — Radioactivity Discovered</h3><p>Henri Becquerel discovers radioactivity.</p><p>Scientists observe mysterious particle emissions.</p><h3>1930 — Neutrino Proposed</h3><p>Wolfgang Pauli proposes the neutrino.</p><p>The missing-energy mystery begins to make sense.</p><h3>1934 — Fermi’s Theory</h3><p>Enrico Fermi develops the first mathematical theory of beta decay.</p><p>This becomes the foundation of weak interaction physics.</p><h3>1956 — Neutrino Detected</h3><p>Clyde Cowan and Frederick Reines finally detect neutrinos experimentally.</p><p>Pauli’s invisible particle is confirmed.</p><h3>1960s–1980s — Electroweak Theory</h3><p>Scientists discover that electromagnetism and the weak force are actually related.</p><p>The theory is developed by:</p><ul><li>Sheldon Glashow</li><li>Abdus Salam</li><li>Steven Weinberg</li></ul><p>This becomes the famous <strong>Electroweak Theory</strong>, one of the greatest achievements in modern physics.</p><h3>1983 — W and Z Bosons Found</h3><p>At CERN, physicists discover the W and Z bosons.</p><p>This provides direct evidence for the theory.</p><p>The discovery earns the Nobel Prize.</p><h3>The Biggest Surprise: The Weak Force Breaks Symmetry</h3><p>For a long time, physicists believed nature behaved identically in a mirror.</p><p>Then came a shock.</p><p>Experiments in 1957 showed the weak force violates mirror symmetry.</p><p>Nature actually distinguishes between left and right.</p><p>No other force behaves this way.</p><p>This discovery stunned the scientific world and led to a Nobel Prize.</p><p>Even today it remains one of the strangest features of fundamental physics.</p><h3>Why Physicists Love the Weak Force</h3><p>The weak interaction is fascinating because it reveals that the universe is not static.</p><p>Matter is constantly capable of transformation.</p><p>Particles are not fixed objects.</p><p>They can change identity under the right conditions.</p><p>The weak force is nature’s mechanism for change itself.</p><p>It drives radioactivity.</p><p>It powers stars.</p><p>It creates neutrinos.</p><p>It helped shape the chemical elements that eventually formed planets and life.</p><p>In many ways, the weak force is the reason the universe evolves rather than remaining frozen forever.</p><h3><strong>What is weak interaction simply?</strong></h3><p><strong>The weak interaction is the fundamental force that allows elementary particles to change type, causing radioactive decay, powering stars, and enabling many of the transformations that shape our universe.</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=396e3e4f6d40" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Cybersecurity in 2026: Security Is Still a Team Sport (But the Rules Have Changed)]]></title>
            <link>https://medium.com/@thinuridulsini/cybersecurity-in-2026-security-is-still-a-team-sport-but-the-rules-have-changed-837bdadfddd6?source=rss-2e65eb35360b------2</link>
            <guid isPermaLink="false">https://medium.com/p/837bdadfddd6</guid>
            <dc:creator><![CDATA[Thinuri Wickramarachchi]]></dc:creator>
            <pubDate>Tue, 23 Jun 2026 05:47:46 GMT</pubDate>
            <atom:updated>2026-06-23T05:47:46.984Z</atom:updated>
            <content:encoded><![CDATA[<blockquote><em>“The attack didn’t start with malware. It started with a meeting invite.”</em></blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*VXCKJb1T8249HE-Dcp8NyA.png" /></figure><p>Monday morning.</p><p>A product manager joins a Teams call that looks legitimate. The invite came from a known partner. The branding is correct. The agenda makes sense.</p><p>Five minutes later, someone shares a document requesting urgent review.</p><p>The document is fake.</p><p>Within hours, credentials are harvested, MFA fatigue prompts begin, and a trusted cloud account becomes an attacker’s foothold.</p><p>Nobody clicked a sketchy attachment.</p><p>Nobody downloaded obvious malware.</p><p>Nobody ignored their annual security awareness training.</p><p>And yet, the organization still got compromised.</p><blockquote>Welcome to cybersecurity in 2026.</blockquote><p>The era where attackers increasingly exploit trust, workflows, identity systems, AI-assisted social engineering, and organizational blind spots rather than relying solely on technical vulnerabilities.</p><p>The lesson?</p><p><strong>Security is still a team sport. But the game has changed.</strong></p><h3>The 2026 Reality | Attackers Target Systems, Not Just Software</h3><p>For years, organizations focused heavily on patching vulnerabilities.</p><p>That remains important.</p><p>But many of the most impactful incidents in recent months have demonstrated something deeper:</p><p>Attackers increasingly target the spaces between teams.</p><p>The handoff between IT and Security.</p><p>The gap between Engineering and Operations.</p><p>The assumption that “someone else owns that.”</p><p>The overlooked SaaS integration.</p><p>The trusted vendor relationship.</p><p>The AI assistant connected to sensitive data.</p><p>The modern attack path often looks less like hacking and more like organizational navigation.</p><h3>Anti-Pattern #1: “Security Is Security’s Job”</h3><p>Imagine a developer saying:</p><blockquote><em>“The security team will catch it.”</em></blockquote><p>Imagine a security analyst saying:</p><blockquote><em>“Infrastructure owns that.”</em></blockquote><p>Imagine an infrastructure engineer saying:</p><blockquote><em>“The vendor manages it.”</em></blockquote><p><strong>This is how risk survives.</strong></p><h3>What Good Looks Like</h3><p>Security ownership becomes distributed.</p><p>Developers understand secure coding.</p><p>Product teams consider abuse cases.</p><p>Infrastructure teams understand identity risks.</p><p>Executives understand operational impact.</p><p>Security teams become enablers rather than gatekeepers.</p><p>The strongest organizations in 2026 are not the ones with the biggest security teams.</p><blockquote>They’re the ones where security knowledge is shared.</blockquote><h3>Best Practice</h3><p>Treat security as a business capability.</p><p>Everyone owns part of the defense.</p><h3>Anti-Pattern</h3><p>Treat security as a department.</p><h3>Anti-Pattern #2: Collecting Security Tools Like Pokémon</h3><p>Many organizations have:</p><ul><li>EDR</li><li>XDR</li><li>SIEM</li><li>SOAR</li><li>CSPM</li><li>DSPM</li><li>CNAPP</li><li>Email Security</li><li>Identity Protection</li><li>Threat Intelligence</li></ul><p>And still struggle to answer:</p><blockquote><em>“Who owns this alert?”</em></blockquote><p>More tools do not automatically create more security.</p><p>In many environments, alert fatigue has become the real threat.</p><h3>What Good Looks Like</h3><p>Security teams prioritize:</p><ul><li>Detection quality over quantity</li><li>Response playbooks over dashboards</li><li>Clear ownership over tool sprawl</li></ul><p>A single high-confidence alert that gets investigated beats 10,000 alerts nobody reads.</p><h3>Best Practice</h3><blockquote>Optimize workflows.</blockquote><h3>Anti-Pattern</h3><blockquote>Optimize vendor count.</blockquote><h3>Anti-Pattern #3: Trusting AI Without Guardrails</h3><p>One of the biggest shifts during 2025 and early 2026 has been the rapid integration of AI into everyday workflows.</p><p>Organizations now use AI for:</p><ul><li>Code generation</li><li>Content creation</li><li>Support operations</li><li>Data analysis</li><li>Research</li><li>Internal assistants</li></ul><p>The problem?</p><blockquote>Many organizations deployed AI faster than they deployed governance.</blockquote><h3>Example</h3><p>A helpful AI assistant is connected to:</p><ul><li>Internal documentation</li><li>HR records</li><li>Customer information</li><li>Engineering systems</li></ul><p>Nobody reviews permissions.</p><p>Nobody maps data exposure.</p><p>Nobody tests abuse scenarios.</p><p>Then an attacker discovers a prompt-injection path.</p><p>The issue isn’t the model.</p><p>The issue is access.</p><h3>Best Practice</h3><p>Treat AI systems like privileged applications.</p><p>Apply:</p><ul><li>Least privilege</li><li>Monitoring</li><li>Data classification</li><li>Security reviews</li></ul><h3>Anti-Pattern</h3><blockquote>Treat AI as “just another productivity tool.”</blockquote><h3>Anti-Pattern #4: MFA Is Enough</h3><p>A few years ago, MFA was considered a silver bullet.</p><p>Today?</p><p>Attackers have evolved.</p><p>Organizations continue facing:</p><ul><li>MFA fatigue attacks</li><li>Adversary-in-the-middle phishing</li><li>Session hijacking</li><li>Token theft</li><li>Identity abuse</li></ul><p>MFA remains critical.</p><p>But it is no longer sufficient by itself.</p><h3>What Good Looks Like</h3><p>Modern identity security includes:</p><ul><li>Phishing-resistant MFA</li><li>Passkeys</li><li>Conditional access</li><li>Device trust</li><li>Session monitoring</li><li>Identity threat detection</li></ul><h3>Best Practice</h3><p>Protect identities continuously.</p><h3>Anti-Pattern</h3><p>Assume login security ends after authentication.</p><h3>Anti-Pattern #5: Security Awareness Once Per Year</h3><p>Every employee remembers the annual training video.</p><p>Very few remember it six months later.</p><p>Attackers know this.</p><p>Modern social engineering is highly contextual.</p><p>Messages reference:</p><ul><li>Real projects</li><li>Real colleagues</li><li>Real vendors</li><li>Real technologies</li><li>Real business events</li></ul><p>Sometimes AI helps attackers personalize at scale.</p><h3>What Good Looks Like</h3><p>Security awareness becomes continuous.</p><p>Small lessons.</p><p>Short simulations.</p><p>Relevant examples.</p><p>Regular discussions.</p><p>Organizations build habits instead of compliance checkboxes.</p><h3>Best Practice</h3><p>Create a culture of reporting.</p><h3>Anti-Pattern</h3><p>Create a culture of blaming.</p><h3>Anti-Pattern #6: Ignoring Third-Party Risk</h3><p>Many breaches no longer begin inside the target organization.</p><p>They begin with:</p><ul><li>Vendors</li><li>Contractors</li><li>SaaS platforms</li><li>Integrations</li><li>Supply chain partners</li></ul><p>Every connection expands trust.</p><p>Every trust relationship expands attack surface.</p><h3>What Good Looks Like</h3><p>Before integrating a vendor, ask:</p><ul><li>What data can they access?</li><li>What happens if they’re compromised?</li><li>How quickly can access be revoked?</li><li>Are permissions reviewed regularly?</li></ul><h3>Best Practice</h3><p>Trust intentionally.</p><h3>Anti-Pattern</h3><p>Trust indefinitely.</p><h3>The Teams That Win in 2026</h3><p>The best cybersecurity teams I’ve seen don’t rely on fear.</p><p>They rely on collaboration.</p><p>Developers ask security questions early.</p><p>SOC analysts talk to engineers.</p><p>Product teams think about abuse cases.</p><p>Leadership understands cyber risk.</p><p>Everyone participates.</p><p>Because security isn’t built during an incident.</p><p>It’s built long before one occurs.</p><p>One conversation.</p><p>One code review.</p><p>One access review.</p><p>One reported phishing email.</p><p>One secure design decision at a time.</p><h3>Final Thought</h3><p>The most dangerous phrase in cybersecurity has never been:</p><blockquote><em>“We’re under attack.”</em></blockquote><p>It’s:</p><blockquote><em>“I thought someone else was handling that.”</em></blockquote><p>Technology matters.</p><p>Tools matter.</p><p>AI matters.</p><p>Threat intelligence matters.</p><p>But in 2026, the organizations that remain resilient understand a timeless truth:</p><blockquote><strong>Security is still a team sport.</strong></blockquote><p>And the strongest defense is not a product.</p><p>It’s a culture where everyone plays their position.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=837bdadfddd6" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[When War Hits Your Phone| The Weekend the Iran–Israel–U.S. Cyber Front Went Civilian]]></title>
            <link>https://medium.com/@thinuridulsini/when-war-hits-your-phone-the-weekend-the-iran-israel-u-s-cyber-front-went-civilian-f237d275b554?source=rss-2e65eb35360b------2</link>
            <guid isPermaLink="false">https://medium.com/p/f237d275b554</guid>
            <category><![CDATA[when-war-hits-your-phone]]></category>
            <dc:creator><![CDATA[Thinuri Wickramarachchi]]></dc:creator>
            <pubDate>Sun, 22 Mar 2026 13:40:03 GMT</pubDate>
            <atom:updated>2026-03-22T13:40:03.063Z</atom:updated>
            <content:encoded><![CDATA[<blockquote>The first sign wasn’t an explosion. It was a notification.</blockquote><p><em>Every named incident below is drawn from public reporting, official statements, or published cybersecurity research. The people and workplaces are fictional composites used to illustrate how these real-world events can feel when they collide at once.</em></p><h3>The morning your dashboards stop telling the truth</h3><p>The first sign isn’t an explosion.</p><p>It’s a message from someone who doesn’t usually panic:</p><blockquote><em>“My calls aren’t going through. Are we under attack?”</em></blockquote><p>In the SOC, you’ve trained yourself not to jump at shadows. Alerts lie all the time. Systems glitch. Users overreact.</p><p>But you also know what modern wars look like now.</p><p>They don’t stay on battlefields. They move through banking systems, logistics networks, telecom providers and the apps people trust without thinking twice.</p><p>You open your monitoring pane expecting the usual: phishing spikes, credential stuffing, noisy scans.</p><p>Instead, you see three things at once:</p><ul><li><strong>Silence</strong> — too much of it. Visibility collapsing where there should be signal.</li><li><strong>Noise</strong> — new domains, strange login patterns, users reporting things that don’t line up.</li><li><strong>Confusion</strong> — screenshots, videos, “proof,” none of it consistently verifiable.</li></ul><p>This is the cyber front in 2026.</p><p>Not a single attack but a convergence of disruption, deception, and psychological pressure, happening faster than attribution… and sometimes faster than basic confirmation.</p><h3>This didn’t start with malware</h3><p>On February 28, 2026, U.S. Central Command announced it had begun <strong>Operation Epic Fury</strong>, with strikes starting at <strong>1:15 a.m. ET</strong>, targeting Iranian command-and-control infrastructure and air defenses.</p><p>But kinetic strikes weren’t the whole story.</p><p>U.S. leadership made something else clear: cyber came first.</p><p>Reporting from Nextgov/FCW described Joint Chiefs Chair Gen. Dan Caine saying <strong>U.S. Cyber Command and U.S. Space Command were “first movers,”</strong> delivering what he called <strong>“non-kinetic effects”</strong> that disrupted communications and sensor networks.</p><p>Before missiles landed, systems were already being shaped.</p><p>And that’s the shift.</p><p>Cyber isn’t support anymore. It’s opening moves.</p><h3>The cameras that watched back</h3><p>For years, “critical infrastructure” meant power plants and pipelines.</p><p>Now it also means traffic cameras.</p><p>According to Financial Times reporting later summarized by The Times of Israel, Israeli intelligence reportedly accessed Tehran’s traffic camera network to track movements tied to senior officials building <strong>pattern-of-life intelligence</strong> from everyday activity.</p><p>The same reporting said cellular service on Tehran’s Pasteur Street was disrupted on the day of an assassination attempt suggesting that <strong>denying communication can matter as much as intercepting it</strong>.</p><p>Even if you treat those claims cautiously, the implication is clear.</p><p>Urban systems the quiet ones are now intelligence surfaces.</p><p>In a SOC, that translates into an uncomfortable checklist:</p><ul><li>Access control systems</li><li>Building management systems</li><li>Logistics tracking</li><li>Vehicle telematics</li><li>Surveillance infrastructure</li></ul><p>Anything that records “normal life” can become operational intelligence.</p><p>And the hardest part?</p><p>If that access was gained months ago, it might have looked like nothing at all.</p><p>Until it doesn’t.</p><h3>When “help is on the way” becomes the attack</h3><p>A few hours later, your phone buzzes.</p><p>A colleague sends a screenshot.</p><p>It’s a push notification.</p><p>It looks official.<br> It feels urgent.<br> It’s exactly the kind of thing people click during a crisis.</p><p>Because ignoring it feels irresponsible.</p><p>In Iran, something like this actually happened.</p><p>WIRED reported that the <strong>BadeSaba Calendar app</strong>, with over <strong>5 million downloads</strong>, appeared to send push notifications urging surrender or defection, promising amnesty. The messages began around <strong>9:52 a.m. Tehran time</strong>, arriving in bursts over about 30 minutes.</p><p>No one claimed responsibility. Attribution remained unclear.</p><p>But the effect was obvious.</p><p>The attacker wasn’t exploiting a vulnerability.</p><p>They were exploiting a moment.</p><p>On the other side of the conflict, a different variation emerged.</p><p>Cybersecurity researchers at Acronis reported a <strong>trojanized version of Israel’s “Red Alert” rocket warning app</strong>, distributed via SMS messages impersonating official communications.</p><p>The app still worked.</p><p>It still delivered alerts.</p><p>But in the background, it harvested:</p><ul><li>SMS messages</li><li>Contacts</li><li>Location data</li><li>Device accounts</li><li>Installed apps</li></ul><p>The Register, citing Acronis, noted something important: <strong>there was no clear way to measure the campaign’s full scope yet.</strong></p><p>Which means the most dangerous attacks might not be the loudest ones.</p><p>They’re the ones that blend in.</p><p>You look at the pattern and realize what’s really being targeted:</p><p>Not your firewall.<br> Not your endpoint detection.</p><p><strong>Your instinct.</strong></p><p>In wartime, people will install anything that promises safety.</p><p>That makes “trusted apps” the most dangerous attack surface on the internet.</p><h3>The internet goes dark</h3><p>At some point, the problem stops being individual incidents.</p><p>It becomes visibility itself.</p><p>On February 28, The Washington Post reported NetBlocks data showing Iran’s internet connectivity dropped to around <strong>4% of normal levels</strong> shortly after the attacks began.</p><p>WIRED reported similar figures, alongside disruptions to phone lines, SMS services, and broadband infrastructure.</p><p>Then the blackout didn’t end.</p><p>NetBlocks later said the shutdown extended into its <strong>20th day</strong>, calling it the <strong>longest recorded internet blackout in Iran’s history</strong>, with only brief, unstable restorations.</p><p>Now your SOC problem changes completely.</p><p>It’s no longer:</p><blockquote><em>“How do we stop the attack?”</em></blockquote><p>It’s:</p><blockquote><em>“How do we operate when we can’t see?”</em></blockquote><p>Because blackouts don’t just silence civilians.</p><p>They reshape reality.</p><p>Fewer videos.<br>Fewer witnesses.<br>Fewer verifiable signals.</p><p>And into that vacuum, something else moves fast.</p><h3>The feed becomes the battlefield</h3><p>You thought defending networks would be the hard part.</p><p>It turns out defending reality is harder.</p><p>The Associated Press reported a <strong>wave of AI-generated and misleading visuals</strong>, some amplified by accounts aligned with state messaging.</p><p>WIRED described how, within minutes of the operation, social platforms filled with:</p><ul><li>Old footage labeled as new</li><li>Incorrect locations</li><li>AI-generated images</li><li>Even video game clips presented as real</li></ul><p>PolitiFact documented similar cases.</p><p>PC Gamer reported that a <strong>War Thunder clip</strong> a video game went viral as supposed real footage and was even shared by a U.S. governor before being deleted.</p><p>Even platforms reacted.</p><p>The Guardian reported that X began restricting monetization for accounts repeatedly posting <strong>unlabeled AI-generated war content</strong>.</p><p>By this point, something becomes impossible to ignore:</p><p>The attack surface has expanded.</p><p>It’s no longer just networks and endpoints.</p><p>It’s <strong>attention</strong>.</p><p>And attention is shaped by algorithms, incentives, and fear.</p><h3>Defending what you can still control</h3><p>Here’s one that doesn’t require guessing, only recognizing patterns already documented.</p><p>If your organization operates in or depends on regions touched by this conflict, your next 30 days shouldn’t focus only on malware prevention.</p><p>They should focus on <strong>resilience to uncertainty</strong>.</p><p>Practical checklist here,</p><ul><li><strong>Maintain an external view of connectivity</strong><br> When your telemetry goes dark, you need independent signals to know whether it’s you or an entire country.</li><li><strong>Treat “trusted apps” as a primary threat vector during crises</strong><br> Both the BadeSaba and Red Alert incidents show that urgency overrides skepticism.</li><li><strong>Assume communications will degrade or be targeted</strong><br> Whether through cyber effects or physical disruption, availability is no longer guaranteed.</li><li><strong>Build misinformation triage into incident response</strong><br> When leadership asks, “Is this real?” your answer shouldn’t depend on social media.</li></ul><p>The lesson of this cyber front isn’t that everything is hackable.</p><p>It’s something more uncomfortable.</p><p>Everything is persuadable.</p><p>Apps.<br>Cameras.<br>Networks.<br>People.</p><p>And if your security strategy only protects the network</p><p>you’re defending yesterday’s perimeter.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f237d275b554" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>