<?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 Em G on Medium]]></title>
        <description><![CDATA[Stories by Em G on Medium]]></description>
        <link>https://medium.com/@gelin2308?source=rss-0ede448f8727------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*Vt3KYBtYjZU1kXG_</url>
            <title>Stories by Em G on Medium</title>
            <link>https://medium.com/@gelin2308?source=rss-0ede448f8727------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sun, 22 Nov 2026 16:52:06 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@gelin2308/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[The dawn of AI Defense: OpenAI unveils Daybreak]]></title>
            <link>https://medium.com/@gelin2308/the-dawn-of-ai-defense-openai-unveils-daybreak-7e4677a229ab?source=rss-0ede448f8727------2</link>
            <guid isPermaLink="false">https://medium.com/p/7e4677a229ab</guid>
            <category><![CDATA[business]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[internet]]></category>
            <category><![CDATA[war]]></category>
            <category><![CDATA[cybersecurity]]></category>
            <dc:creator><![CDATA[Em G]]></dc:creator>
            <pubDate>Fri, 15 May 2026 13:11:25 GMT</pubDate>
            <atom:updated>2026-05-15T13:11:25.547Z</atom:updated>
            <content:encoded><![CDATA[<p>Like Anthropic’s Mythos,</p><p>OpenAI launched Daybreak on May 12, 2026, a cybersecurity initiative combining frontier AI models with Codex Security.</p><p>“The goal is simple : accelerate cyber defenders and continuously secure software”, argues OpenAI on his website.</p><h3>Relies on GPT — 5.5</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*TePknpB69CbsesgF7uGE8A.png" /></figure><p>Daybreak helps organizations identify and patch vulnerabilities before exploitation. It integrates secure code review, threat modeling and patch validation into development loops. This tool uses three distinct models:</p><ul><li>GPT-5.5 for general use</li><li>GPT-5.5 with Trusted Access for Cyber for defensive tasks</li><li>GPT-5.5-Cyber for red teaming.</li></ul><p>By leveraging AI as an agentic harness, OpenAI aims to tilt the security balance toward defenders, allowing them to address issues in isolated environments and propose fixes for high-impact code and attack paths.</p><p>The rollout addresses a shift where AI-assisted research has dramatically increased vulnerability discovery speeds, causing triage fatigue. Major firms like Cisco and Cloudflare are already integrating these capabilities.</p><p>In March, HackerOne paused its bug bounty program due to this volume surge. Security researcher Himanshu Anand noted that the traditional 90-day disclosure window fails, as AI can now convert patch diffs into working exploits in just 30 minutes. Consequently, companies like OpenAI and Anthropic deploys AI security agents as critical operational layers to safeguard digital infrastructure and manage the remediation bottleneck created by near-zero exploit timelines.</p><h3>Defenders first ?</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*6y-cyajkMk58HfquM-UK5g.jpeg" /></figure><p>While similar AI capabilities have existed, Daybreak’s significance lies in its strategic market positioning and robust technical functions.</p><p>Technically, Daybreak:</p><ul><li>Analyzes a code repository to produce an editable threat model</li><li>Runs candidate vulnerabilities in an isolated environment for safe testing</li><li>Proposes and validates fixes for identified issues.</li></ul><p>Its launch is anchored by major partners, such as Akamai, Cisco, Cloudflare, CrowdStrike, Fortinet, Oracle, Palo Alto Networks or Zscaler, positioning OpenAI as the premier AI partner integrated directly into the established security stack to accelerate defenders.</p><h3>Mythos : an offensive stance ?</h3><p>Anthropic’s Mythos was introduced with an offensive framing, contrasting OpenAI’s Daybreak. Rather than launching a defensive product, Mythos was unveiled through a research disclosure demonstrating Claude’s capability to discover, link and exploit zero-day vulnerabilities at machine speed.</p><p><strong>Technical Capabilities for Business:</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/749/1*dzuTghCC3nOSCeJX2DnuOQ.png" /></figure><p>Anthropic leveraged this “what it could do to you” pitch for private engagement with major companies and federal agencies, positioning the offensive capability alongside a responsible release strategy. Despite the different market posture, the underlying AI security technology is similar to Daybreak’s.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7e4677a229ab" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Myth of Mythos: Why the AI Cyber War is already here ?]]></title>
            <link>https://medium.com/@gelin2308/the-myth-of-mythos-why-the-ai-cyber-war-is-already-here-8b0f993543c2?source=rss-0ede448f8727------2</link>
            <guid isPermaLink="false">https://medium.com/p/8b0f993543c2</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[war]]></category>
            <category><![CDATA[anthropic-claude]]></category>
            <category><![CDATA[business]]></category>
            <category><![CDATA[openai]]></category>
            <dc:creator><![CDATA[Em G]]></dc:creator>
            <pubDate>Sat, 09 May 2026 19:52:52 GMT</pubDate>
            <atom:updated>2026-05-09T19:52:52.930Z</atom:updated>
            <content:encoded><![CDATA[<p>Users accessed Claude Mythos Preview via a contractor account.</p><p><strong>Mythos scares everyone ?</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*3aoFyb3O6ZwEm6qrs2PVyA.jpeg" /></figure><p>Anthropic’s release of the Mythos model has caused significant concern among global banks, tech giants and governments by identifying thousands of previously unknown software vulnerabilities. To mitigate risks, Anthropic limited its initial rollout to a few major American companies, including Apple, Amazon, JPMorgan Chase and Palo Alto Networks, as part of a security measure called Project Glasswing.</p><p>Despite these precautions, the Trump administration considers new oversight for future AI models. For Anthropic CEO Dario Amodei, he warned of an enormous increase in breaches and financial damage from ransomware targeting schools, hospitals and banks as AI capabilities rapidly advance.</p><p>However, cybersecurity experts and researchers argue that the capabilities demonstrated by Mythos are already achievable using existing, older models through clever orchestration.</p><p>Firms like Vidoc and Aisle successfully reproduced Mythos’s results using older models from both Anthropic and OpenAI. Anthropic previously warned of this trend, noting that its Claude Opus 4.6 model identified over 500 high-severity vulnerabilities in open-source software. The primary advantage currently lies with offensive actors, as corporations often take days or weeks to patch systems. This widening gap lowers the barrier for criminal groups and adversarial nations to automate exploits with minimal human input.</p><p><strong>Race for AI Supremacy</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*K_g7-lYCEvFxTI1_vW4K5w.png" /></figure><p>Despite controversy, Anthropic reports in discussions to raise up to $50 billion in a funding round that could value the AI startup at $900 billion. This valuation would surpass OpenAI, which reached an $852 billion post-money valuation following its $122 billion financing round in March 2026.</p><p>The capital intends to expand Anthropic’s computing capacity to support its Claude model family, particularly the new cybersecurity-focused Claude Mythos. Currently, Anthropic’s annualized revenue run rate has surpassed $30 billion, a significant jump from $9 billion at the end of 2025, driven by enterprise adoption and products like Claude Code.</p><p>The competitive landscape between Anthropic and OpenAI intensifies as both companies eye potential initial public offerings. While OpenAI CEO Sam Altman aims for a listing as early as late 2026, CFO Sarah Friar has suggested 2027 to meet rigorous financial standards.</p><p>OpenAI currently generates approximately $25 billion in annualized revenue but faces a $1.15 trillion infrastructure obligation through 2030. In contrast, Anthropic demonstrates higher revenue efficiency, producing $6 million per employee compared to OpenAI’s $5.6 million. If Anthropic completes this $900 billion round, it would establish a formidable valuation benchmark before transitioning to public markets.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8b0f993543c2" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Anthropic Launches Claude Security: AI-Driven defense for the enterprise]]></title>
            <link>https://medium.com/@gelin2308/anthropic-launches-claude-security-ai-driven-defense-for-the-enterprise-a13f7edfd1bc?source=rss-0ede448f8727------2</link>
            <guid isPermaLink="false">https://medium.com/p/a13f7edfd1bc</guid>
            <category><![CDATA[security]]></category>
            <category><![CDATA[cybersecurity]]></category>
            <category><![CDATA[business]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[internet]]></category>
            <dc:creator><![CDATA[Em G]]></dc:creator>
            <pubDate>Fri, 01 May 2026 21:07:19 GMT</pubDate>
            <atom:updated>2026-05-01T21:07:19.414Z</atom:updated>
            <content:encoded><![CDATA[<p>Anthropic employs a security methodology that differs fundamentally from that of the U.S. government.</p><p>During the last week of April, Anthropic launched Project Glasswing, a cybersecurity initiative involving major tech leaders like Amazon Web Services, Google and Microsoft. The project utilises Claude Mythos Preview, a frontier AI model with advanced coding capabilities.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*KQ86MBADM-wArTxG6vcnCQ.jpeg" /></figure><p><strong>Mythos Preview in action</strong></p><p>Mythos Preview has already identified thousands of high-severity flaws in major operating systems and browsers, including a 27-year-old OpenBSD vulnerability. Anthropic committs $100 million in usage credits and $4 million in donations to bolster defensive security. This collaboration aims to give cyber defenders a durable advantage by identifying and fixing critical bugs before they can be exploited.</p><p>To support global infrastructure, the Project Glasswing provides over 40 organisations with access to Mythos Preview for securing first-party and open-source systems. Evaluation benchmarks like CyberGym show the model achieves an 83.1% reproduction rate, significantly outperforming Claude Opus 4.6. Partners use these capabilities for local vulnerability detection, penetration testing and securing endpoints.</p><p>Anthropic plans to share industry-wide findings within 90 days and collaborate with governments to mitigate national security risks. By automating triage and patching processes, the initiative seeks to modernise cybersecurity stacks. This effort reflects a critical shift toward using AI for proactive defense against increasingly sophisticated and frequent cyberattacks.</p><p><strong>The risks with Mythos Preview</strong></p><p>On April 7, 2026, researchers including Nicholas Carlini, Newton Cheng and Keane Lucas published an evaluation of the Claude Mythos Preview model’s cybersecurity capabilities. This report, found at red.anthropic.com, explores the potential risks associated with the model’s deployment in sensitive digital environments.</p><p>The study specifically focused on assessing the model’s proficiency in vulnerability discovery, exploit generation and social engineering. Through rigorous red-teaming exercises, the team aimed to determine if the model significantly lowers the barrier for cyberattacks. This assessment serves as a foundational benchmark for Anthropic, ensuring that technological advancement balances with rigorous safety protocols and transparency.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*LCvohtPk7J-AP-UAS6Qdyg.jpeg" /></figure><p>The findings suggest that Claude Mythos Preview demonstrates advanced reasoning while remaining within safe operational parameters for cybersecurity. The evaluation tested the model against numerous benchmarks to verify that it does not provide substantial assistance for sophisticated malicious activities. Data from the report illustrates the model’s performance across multiple categories, confirming that safety fine-tuning effectively mitigates the risk of generating actionable cyberattack content. The researchers emphasize that ongoing monitoring and proactive updates should stay ahead of evolving threats. Ultimately, the report highlights the importance of integrating robust defensive layers to protect digital infrastructure against potential future AI misuse.</p><p><strong>Claude Code Security on the web</strong></p><p>In the middle of April 2026, Anthropic launched Claude Security in public beta for Enterprise customers, providing a tool designed to scan codebases and suggest fixes. Powered by the Claude Opus 4.7 model, the platform moves beyond simple pattern matching.</p><p>Built into Claude Code on the web, it reasons through code like a human security researcher, understanding complex file interactions and data flows. Formerly known as Claude Code Security, it was tested by hundreds of organisations before this broader rollout. Accessible through Claude.ai, it requires no complex API setup, lowering barriers for engineering teams seeking advanced AI-driven security solutions.</p><p>Recently, Anthropic tries to establish a strong partner network with technology giants like CrowdStrike, Microsoft Security and Palo Alto Networks integrating Opus 4.7 into their platforms. Additionally, global services firms including Accenture, Deloitte and PwC deploy Claude-integrated solutions for clients.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a13f7edfd1bc" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Broadcom and Meta : A Multi-Year AI Power Play to 2029]]></title>
            <link>https://medium.com/@gelin2308/broadcom-and-meta-a-multi-year-ai-power-play-to-2029-0f771a9f9d54?source=rss-0ede448f8727------2</link>
            <guid isPermaLink="false">https://medium.com/p/0f771a9f9d54</guid>
            <category><![CDATA[business]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[google]]></category>
            <category><![CDATA[metaverse]]></category>
            <dc:creator><![CDATA[Em G]]></dc:creator>
            <pubDate>Sat, 25 Apr 2026 20:04:28 GMT</pubDate>
            <atom:updated>2026-04-25T20:15:30.129Z</atom:updated>
            <content:encoded><![CDATA[<p>AI chip business grows while Meta counts every click</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ScKZQf6kvmiQzlWGEgJcHA.png" /></figure><p>Broadcom (AVGO) demonstrates a remarkable acceleration in its AI semiconductor business, positioning itself as a critical enabler in the current technological boom.</p><p>Let me explain !</p><p>Broadcom’s financial performance reflects ascendance, with AI semiconductor revenue climbing 106% year-over-year to $8.4 billion in Q1 2026. This significant jump comes from the insatiable demand out of hyperscale cloud providers and AI leaders, including key customers such as Anthropic, Fujitsu and OpenAI. Analysts even forecast a potential path to achieving over $100 billion in AI chip revenue by 2027, underscoring the market’s confidence in Broadcom’s strategic focus and execution capabilities within this high-growth sector.</p><p>While the rapid influx of AI-related revenue led to upward revisions of overall AVGO estimations, it introduces a dynamic market complexity. The current mix of AI revenue weakens the company’s gross margin, a common effect when aggressively scaling production and fulfilling high-volume, customized orders for concentrated customer bases. However, this immediate margin pressure counterbalances a strong operating leverage, which projects to bolster operating profit and earnings per share (EPS) growth. Broadcom’s performance illustrates the dual nature of hyper-growth in the semiconductor industry: immense top-line expansion coupled with the challenges of managing margin volatility and substantial supply chain execution risk.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FIH2bfD_LhYo%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DIH2bfD_LhYo&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FIH2bfD_LhYo%2Fhqdefault.jpg&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/3d9707718ca3786dbe05fd60bffcbc86/href">https://medium.com/media/3d9707718ca3786dbe05fd60bffcbc86/href</a></iframe><p><strong>A multi-generation strategy !</strong></p><p>Broadcom and Meta Platforms recently announced multi-year, multi-generation strategic partnership especially for dominance in the foundational AI infrastructure space.</p><p>This alliance, set to extend through at least 2029, centers on Broadcom’s commitment to co-develop multiple generations of Meta’s custom AI silicon, the Meta Training and Inference Accelerator (MTIA) chips, which are critical for powering AI across Meta’s vast ecosystem of apps and services. The scale comes immediate and massive, with an initial commitment exceeding one gigawatt, kicking off a planned multi-gigawatt deployment over the next three years. This deep collaboration solidifies Broadcom’s position as a core technology provider for a hyperscale AI leader.</p><p>Also, Broadcom wants to extend its portfolio of industry-leading networking solutions to provide key components such as:</p><ul><li>High-radix Ethernet switches</li><li>Optical Connectivity products</li><li>PCIe switches</li><li>High-speed SerDes capabilities.</li></ul><p>Broadcom CEO Hock Tan transitioned from Meta’s board to an advisory role focused on their custom silicon roadmap. This move ensures Broadcom’s technological expertise directly influences the future shape of Meta’s infrastructure investments, confirming the alliance as a fundamental component of both companies’ long-term AI strategy.</p><p>Meta plan data-gathering</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FrXSPopXet1o%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DrXSPopXet1o&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FrXSPopXet1o%2Fhqdefault.jpg&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/f88a1524a6a71eff2b85f75e8cda8998/href">https://medium.com/media/f88a1524a6a71eff2b85f75e8cda8998/href</a></iframe><p>Meta currently prepares a massive internal data-gathering project called the Model Capability Initiative (MCI) to accelerate the training of its artificial intelligence models. This initiative wants to create a “big and unbiased” dataset reflecting how employees perform tasks on their corporate devices. The primary method involves an internal tracking tool that captures employee inputs, such as keystrokes, mouse clicks and on-screen content, across hundreds of popular third-party applications and sites. This includes Google, LinkedIn, Wikipedia and developer tools like GitHub and Slack. The core goal consists to develop sophisticated AI agents capable of performing complex, real-world white-collar tasks, such as office and coding-related functions.</p><p>This data-gathering push combines the CEO Mark Zuckerberg’s ambitious strategy to close the gap with generative AI leaders like OpenAI and Google, following a recent spending spree and team-building effort led by Scale AI’s Alexandr Wang. The company recently debuted a new foundation model series, Muse Spark, developed by its MSL unit. However, the MCI tool has sparked significant controversy internally, with multiple employees characterizing the project as “dystopian” and expressing major concerns over privacy and the risk of widely exposing sensitive data, including personal, immigration or new product development details. Meta’s assurances confirm the data remains solely for model training.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=0f771a9f9d54" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The AI Revolution foresees an Industrial Era Boom]]></title>
            <link>https://medium.com/@gelin2308/the-ai-revolution-foresees-an-industrial-era-boom-6b904264e510?source=rss-0ede448f8727------2</link>
            <guid isPermaLink="false">https://medium.com/p/6b904264e510</guid>
            <category><![CDATA[future]]></category>
            <category><![CDATA[business]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[productivity]]></category>
            <category><![CDATA[economy]]></category>
            <dc:creator><![CDATA[Em G]]></dc:creator>
            <pubDate>Thu, 16 Apr 2026 13:09:11 GMT</pubDate>
            <atom:updated>2026-04-16T13:09:11.041Z</atom:updated>
            <content:encoded><![CDATA[<p>“AI is coming, are we going to lead it ?”</p><p>Phil Gramm, professor at Texas A&amp;M University, talked about the AI revolution during the Weidenbaum Center on the Economy, Government and Public Policy event last Thursday.</p><p>He thinks it could be “Drammatic”.</p><p><strong>New Industrial Revolution ahead</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*fauH8LMpmpZpp8RzhfELaQ.png" /></figure><p>Former Senator and economist <strong>Phil Gramm</strong> recently delivered a lecture at WashU, sponsored by the Weidenbaum Center, where he positioned artificial intelligence as a primary driver of future economic prosperity.</p><p>Central to his argument involves a reevaluation of the Industrial Revolution, which he claims is unfairly maligned by historical myths and literary fiction. Gramm asserted that this era actually triggered unparalleled improvements in wages, literacy and general public well-being. By debunking the notion that industrialization only enriched elites, he established a historical precedent for technological shifts creating massive societal gains rather than widespread impoverishment or lasting urban decay.</p><p>Drawing direct parallels to the modern era, he argued also that current anxieties regarding AI mirror the unfounded fears of the nineteenth century. He warned that government intervention and regulatory hurdles could stunt innovation, potentially allowing global competitors to surpass the United States.</p><p>While acknowledging current technological limitations, Gramm remains optimistic that AI will ultimately make humanity richer and freer. However, he also voiced concerns over contemporary fiscal policies, criticizing protective tariffs and rising national debt, ultimately urging the audience to embrace AI and financial literacy to ensure America leads this next great technological revolution.</p><p><strong>PwC confirms the trends</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jWHcHDSi4VpYCo1ODyXwTQ.png" /></figure><p>A recent PwC study reveals a significant economic divide in the corporate world, with a mere 20% of organizations capturing 74% of the total financial value generated by artificial intelligence.</p><p>While many firms remain stagnant in the pilot phase, these “AI leaders” successfully convert technological experimentation into measurable returns. The research, which surveyed over 1,200 senior executives globally, indicates that top performers distinguish themselves by utilizing AI as a tool for total business reinvention. Rather than focusing solely on internal efficiencies, these companies leverage AI to explore new revenue streams and capitalize on opportunities where different industries converge.</p><p>The study further highlights that superior financial outcomes are driven by advanced automation and robust governance frameworks. AI leaders are nearly three times more likely than their peers to automate decision-making processes and twice as likely to utilize autonomous, self-optimizing systems. This high level of integration is supported by a commitment to “trust at scale,” involving responsible AI frameworks and cross-functional oversight boards. Consequently, employees at these leading firms report significantly higher trust in AI outputs. Without a strategic shift toward growth and scalable governance, the performance gap between these pioneers and lagging businesses expects to widen.</p><p><strong>The Industry 4.0 core</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_7fBT_uKiUhBww-lspf-iw.jpeg" /></figure><p>Industry 4.0, or smart manufacturing, represents the fourth industrial revolution by integrating digital technologies into the factory floor to revolutionize production.</p><p>Building upon previous eras of steam, electricity and basic computing, this modern phase utilizes the Internet of Things (IoT), cloud computing and AI to create interconnected systems. By embedding advanced sensors and robotics into production lines, manufacturers achieve real-time visibility and data-driven decision-making.</p><p>This digital transformation eliminates traditional information silos, allowing for seamless integration between factory operations and enterprise systems like supply chains. Consequently, companies can achieve higher productivity, improved yields and greater organizational agility.</p><p>By analyzing big data through machine learning, manufacturers can also detect defects early and simulate production changes virtually to minimize downtime. Furthermore, edge computing ensures rapid responses for safety and quality control by processing data locally, while robust cybersecurity protects these new digital pathways. This shift enables “mass customization,” where factories can efficiently produce specialized items at a low cost.</p><p>Ultimately, Industry 4.0 empowers businesses to be more responsive to customer demands while significantly reducing waste and operational costs.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6b904264e510" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[AI Agents kill the Search Box]]></title>
            <link>https://medium.com/@gelin2308/ai-agents-kill-the-search-box-00cdf6736703?source=rss-0ede448f8727------2</link>
            <guid isPermaLink="false">https://medium.com/p/00cdf6736703</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[seo]]></category>
            <category><![CDATA[website]]></category>
            <category><![CDATA[google]]></category>
            <category><![CDATA[business]]></category>
            <dc:creator><![CDATA[Em G]]></dc:creator>
            <pubDate>Fri, 10 Apr 2026 13:13:19 GMT</pubDate>
            <atom:updated>2026-04-10T13:13:19.549Z</atom:updated>
            <content:encoded><![CDATA[<p>AI chooses the result in the place of the user !</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*tvJHOEyWMgZckRs4bzk7TA.png" /></figure><p>Among Google CEO Sundar Pichai, the future of Search in the age of advanced AI, evolved dramatically rather than being replaced. “<em>Information seeking queries will be agentic search</em>”, he says in the Search Engine Journal.</p><p><strong>No search box needed</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*nvBEUP2wN_MFHkqdIyXM0g.png" /></figure><p>AI agents complete multi-step tasks and handle numerous asynchronous threads for the user. Essentially, Search will transform from a distribution mechanism for web links into an orchestration layer that manages these various AI agents to accomplish complex user goals. He avoids the idea of obsolescence, viewing the shift as an expansionary moment for the product frontier.</p><p>Sundar Pichai views the change not as a zero-sum game but as an expansion of capabilities, emphasising rapid evolution over long-term predictions due to the steep curve of AI advancement. Practically, in ten years, suggesting Search will function as an “agent manager” facilitating complex, long-running tasks. This vision significantly abstracts the user’s interaction away from traditional ranked results, moving toward comprehensive task completion. The reason is that Google must embrace competitive agility and the fusion of Search and Gemini to remain at the cutting edge of this technological shift.</p><p><strong>No website at all ?</strong></p><p>A critical point of concern for the publishing and SEO communities remains the conspicuous absence of “websites” or “web pages” in the future search. Pichai mentioned only briefly as data points to be summarized or understood.</p><p>This omission, in a conversation about the world’s leading search engine, underscores the implicit shift toward treating the web’s content as raw data for AI agents to consume and synthesize, rather than as destinations for user clicks. Thus, while Google may thrive by embracing this “agentic” future, businesses relying on web traffic for monetisation must urgently adopt competitive agility to avoid becoming irrelevant in a landscape where the traditional Search Box will increasingly be bypassed.</p><p><strong>“Near me” become “Ask Maps”</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*yW2OUUiVUE-YwjrXSvS62w.png" /></figure><p>This year (2026), the most important impact of AI Agent in Search Engines concerns Google Maps.</p><p>It undergoes a major transformation, leveraging the power of Gemini models to fundamentally change how users explore and navigate. This reimagining introduces <strong>Ask Maps</strong>, a new conversational AI experience that moves beyond simple location searches to answer complex, real-world queries.</p><p>Users can now ask sophisticated questions, such as finding a public tennis court with lights or a convenient, mid-way vegan restaurant for a group, and receive :</p><ul><li>Personalized recommendations</li><li>Custom maps</li><li>Actionable steps like booking reservations.</li></ul><p>Ask Maps analyzes vast amounts of data, including over 300 million places and contributions from over 500 million users, to provide insider tips and relevant results, rolling out now in the U.S. and India.</p><p>In addition to enhanced exploration, Google Maps also launches <strong>Immersive Navigation</strong>, its biggest driving update in over a decade, designed to make trips more intuitive and stress-free.</p><p>This feature brings the map to life with vivid 3D views that accurately reflect buildings, overpasses and terrain. It uses Gemini to analyze Street View and aerial imagery for spatial understanding. New functionality includes intelligent voice guidance, a broader view of the route with smart zooms to prepare for tricky maneuvers and real-time alerts about road conditions for example.</p><p>Immersive Navigation also assists with the final stretch, providing Street View previews, parking recommendations and entrance highlighting.</p><p><strong>This is AEO</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*sOYiJgReERg3dzJyW5AGIA.jpeg" /></figure><p>The rise of AI-powered search causes also a massive shift for businesses, exemplified by HubSpot losing 140 million annual visits as users increasingly bypass traditional web links.</p><p>This decline could be fueled by search engines’ AI overviews, which reduce click-through rates by 60 % to 70 %. Companies now urgently adopt <strong>Answer Engine Optimisation (AEO)</strong>, also known as Generative Engine Optimisation (GEO), to ensure their content appears within AI-generated answers. This strategic change requires businesses to move beyond traditional SEO and adapt their content to the new, more specific, and long-form conversational queries typical of AI search, which can be 40 to 60 words long compared to 4 to 6 words in a traditional search.</p><p>Successful AEO involves restructuring content into:</p><ul><li>Smaller</li><li>Easily extractable “chunks”</li><li>Focusing on demonstrating expertise and authority (E-A-T).</li></ul><p>For example, HubSpot now uses shorter content pieces, while Spice Kitchen is building authority-demonstrating content clusters on the history of the spice trade, viewing their website as an educational resource rather than just a shop. Businesses like MKM Building Supplies are creating content with clear summaries, bulleted lists and FAQs to make it easily digestible for Large Language Models (LLMs). Although AI drives fewer initial visitors, these users are highly qualified, with MKM reporting that AI-influenced visitors are significantly more likely to convert into paying customers.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=00cdf6736703" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[ChatGPT’s shifting landscape : why users are seeking alternatives ?]]></title>
            <link>https://medium.com/@gelin2308/chatgpts-shifting-landscape-why-users-are-seeking-alternatives-745d264a473d?source=rss-0ede448f8727------2</link>
            <guid isPermaLink="false">https://medium.com/p/745d264a473d</guid>
            <category><![CDATA[google]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[openai]]></category>
            <category><![CDATA[finance]]></category>
            <category><![CDATA[data-science]]></category>
            <dc:creator><![CDATA[Em G]]></dc:creator>
            <pubDate>Fri, 03 Apr 2026 08:27:17 GMT</pubDate>
            <atom:updated>2026-04-03T08:27:17.102Z</atom:updated>
            <content:encoded><![CDATA[<p>Users talk about quality regression.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*m2IVGaLM1b1eQsuzIICt9Q.png" /></figure><p>I rarely use ChatGPT but users qualify the tool as not very intelligent at all lately !</p><p>The consensus among users declares clearly : ChatGPT’s quality dropped noticeably in 2026. This perception anchors itself in significant technical shifts at OpenAI.</p><p><strong>May due to GPT-5</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*7FzbiaVcF-IFltO6JjoMgg.png" /></figure><p>The core issue arises from the transition from the highly-praised GPT-4 to the newer, cost-optimised GPT-5.x model family. While GPT-5.x performs excellently on standardised benchmarks for reasoning and math, it exchanges the detailed, helpful and nuanced responses of its predecessor for faster inference and stringent safety guardrails.</p><p>User complaints suggest : lazier responses, shrinking outputs, increased refusals, triggering safety flags on benign creative or hypothetical queries. This “alignment tax” prioritising benchmark scores and safety over real-world user satisfaction combines with cost-optimised inference routing, which sends simple queries to lighter, less capable model variants.</p><p><strong>Valuable alternative</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*iq7ph2f7QhQT_Xg5UI9_ig.jpeg" /></figure><p>While OpenAI continues to iterate, the structural pressures of scale and safety mean users must now actively choose the best tool for their specific workflow, with Claude currently leading the pack for general-purpose high-quality output.</p><p>Anthropic’s <strong>Claude</strong> has emerged as the primary competitor, particularly among developers, achieving 43% adoption in the 2025 Stack Overflow survey. Claude Opus 4.6 excels where ChatGPT falters :</p><ul><li>producing superior writing quality</li><li>following complex instructions faithfully</li><li>consistently generating complete code implementations.</li></ul><p>For specific needs, <strong>Gemini 3.1 Pro</strong> dominates multimodal and research tasks with its 2M token context window, while <strong>Perplexity</strong> could be unmatched for research requiring real-time web citations. For users focused on quality and budget, <strong>DeepSeek V4</strong> offers near-frontier model performance at a fraction of the cost.</p><p><strong>The race to the public market</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Ubg-NwedSDpsTFbQSYnO0g.png" /></figure><p>The true blockbuster of 2026 presents the intense <strong>Race to the Public Market</strong> among AI heavyweights, chiefly OpenAI and Anthropic.</p><p>OpenAI’s colossal $122 billion private financing, which provided it an aggressive $852 billion valuation, essentially purchases <strong>time</strong> to position itself for a successful IPO that can absorb a valuation priced on its performance three to four years out. The complex financing, involving commitments from Amazon, NVIDIA and Microsoft, largely circulates with funds flowing back to the investors as cloud and hardware purchases.</p><p>This massive capital injection requires urgent coverage for astonishing burn rates. OpenAI projects losing $14 billion in 2026, suggesting the $122 billion may only last 18 to 24 months. The public market represents the last major untapped source of capital to sustain operations.</p><p>Rival Anthropic complicates this goal. Valued at $380 billion and rapidly gaining ground in the enterprise API market, Anthropic widely predicts going public before OpenAI. If Anthropic consumes a substantial portion of the available investor capital, it could put downward pressure on OpenAI’s eventual IPO pricing. Ultimately, all the financing and IPO planning hinge on whether AI companies can generate revenue faster than they incur expenses.</p><p>Anyway, the company that goes public first may secure its survival.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=745d264a473d" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The AI Generation Gap: What kids are hiding about AI ?]]></title>
            <link>https://medium.com/@gelin2308/the-ai-generation-gap-what-kids-are-hiding-about-ai-cb8aa3e83709?source=rss-0ede448f8727------2</link>
            <guid isPermaLink="false">https://medium.com/p/cb8aa3e83709</guid>
            <category><![CDATA[parents]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[children]]></category>
            <category><![CDATA[education]]></category>
            <category><![CDATA[chatbots]]></category>
            <dc:creator><![CDATA[Em G]]></dc:creator>
            <pubDate>Thu, 26 Mar 2026 13:27:03 GMT</pubDate>
            <atom:updated>2026-03-26T13:27:03.166Z</atom:updated>
            <content:encoded><![CDATA[<p>Parents think they know how kids use AI. They don’t.</p><p>There’s a serious lack of communication about AI within families, says Monica Anderson, managing director at the Pew Research Center, based on a survey of 1,458 American teens aged 13–17 and their parents.</p><p>Let me show you !</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*UEo8AGHaxzmmlBfM96drkA.png" /></figure><h3>Commonly</h3><p>Teenagers are less worried about AI than their parents, but many adults are oblivious to what kids are doing with AI in the first place.</p><p>In fact, AI plays a huge role in Isis Joseph’s life. “I use it every day,” she says. The 17-year-old 11th grade student in New York City uses it for homework. AI helps her decide where to eat and provides inspiration for her poetry. Sometimes, she even turns to it with questions about issues in her personal life. “Parents may inflate AI as this very threatening thing”, and of course, many of those concerns remain valid, she says. “But I think AI is generally good.”</p><p>For parents, when Pew asked them if their children use AI, only 51% said yes. The truth is 64% of teenagers said they use chatbots. Common Sense Media found similarly radical differences. Millions of parents are in the dark about what happens on their kids’ screens. That makes sense, because according to Pew, four out of 10 parents said they have never had a conversation with their children about AI.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*7qR92zleFaTmilFzS9feoA.jpeg" /></figure><h3>Emotional support</h3><p>When studies compared what children were doing with their parents’ expectations, they found significant differences. It seems like a lot of teens are making decisions about AI on their own.</p><p>Of all parents’ anxieties about what their children are doing with chatbots, companionship stood out. According to Pew, 58% of American parents said they’re not okay with their teens using AI for emotional support and another 20% said they weren’t sure. But it’s happening.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*K3wGXg0gwnlJuPnuv7ZrYg.jpeg" /></figure><h3>Red Flags</h3><p>According to studies, the signs of problematic AI use in teens may include :</p><ul><li>A new “best friend” or primary confidant</li><li>A fall apart feeling when AI remains inaccessible</li><li>School, sleep or real friendships are slipping</li><li>Noticeable changes in mood, behaviour or thinking.</li></ul><p>When Pew asked US teenagers about AI, 12% said they use it for advice or emotional support and 16% said they use it for casual conversation. Those proportions may be small, but they still amount to millions of children across the US if the survey is representative.</p><h3>Work and play</h3><p>Among teens, some of the most common AI habits are what you might expect. “I usually use it to study,” says Eloise Chu, a 13-year-old from Chatham, New Jersey in the US. “Like if I have a math test, I’ll give it a problem I don’t know how to do so it can generate more questions, so I can get good at it.”</p><p>But experts warn a lack of communication means many teens are left to navigate AI without parental guidance.</p><p>However, about other students, there is a very different picture — 59% of teenagers told Pew that students at their school use AI to cheat while 34% said it happens extremely or very often.</p><p>Sometimes, there are classmates who literally yell at teachers,”Hey, if you don’t come answer my question, I’m just gonna get AI to do this for me” student says. A 14-year-old boy, has similar stories. “In science class, we had to research a topic and write about it and one of the kids at my table had just completely copied what the AI told him,” he says. “But then he couldn’t read his own handwriting and he didn’t even remember what he wrote.”</p><h3>Attitude problems</h3><p>According to a study from Common Sense Media, 52% of parents say using AI in school assignments is “unethical and should have consequences”. But ask the teenagers, and the exact same number say using AI for schoolwork is “innovative and should be encouraged”. Either the children are missing something, or the parents are.</p><p>But teens seem more comfortable with the tools in general. Common Sense Media found that 92% of teens say they can tell when they’re interacting with an AI system or a real human, compared to 73% of parents. According to Pew, almost six out of 10 teenagers say they’re confident about their ability to use chatbots, and a quarter said they’re very or extremely confident.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=cb8aa3e83709" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Trump’s AI Policy Explained: The National Framework for American AI Dominance]]></title>
            <link>https://medium.com/@gelin2308/trumps-ai-policy-explained-the-national-framework-for-american-ai-dominance-726087f6d10f?source=rss-0ede448f8727------2</link>
            <guid isPermaLink="false">https://medium.com/p/726087f6d10f</guid>
            <category><![CDATA[children]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[war]]></category>
            <category><![CDATA[donald-trump]]></category>
            <category><![CDATA[law]]></category>
            <dc:creator><![CDATA[Em G]]></dc:creator>
            <pubDate>Fri, 20 Mar 2026 21:13:31 GMT</pubDate>
            <atom:updated>2026-03-20T21:13:31.928Z</atom:updated>
            <content:encoded><![CDATA[<p>The Trump Administration asserts its commitment to securing American dominance in the artificial intelligence sector.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*1bvFIwqdYlKno24sJIiohg.png" /></figure><h3>The New National Framework for AI</h3><p>On 20th of March 2026, The White House aims to initiate an era of economic supremacy, enhanced national security and widespread human prosperity. Achieving this objective necessitates a consistent, national policy structure that simultaneously fosters American industry innovation and guarantees equitable benefits for the populace. Recognizing public apprehension regarding AI’s impact on children’s safety and energy costs, the Administration demonstrates decisive leadership by presenting a comprehensive national legislative framework. This mandate comprises seven core policy directions designed to supersede disparate state-level regulations and cement Federal authority over AI development :</p><ol><li><strong>Protecting children and empowering parents, </strong>Congress must furnish robust tools, such as account controls, to manage children’s digital exposure and safeguard privacy. Also, AI platforms accessible by minors require implementing features that actively reduce the risk of sexual exploitation and discourage self-harm behaviors.</li><li><strong>Safeguarding and strengthening American communities,</strong> AI progress must stimulate every organism through economic expansion and energy superiority. The Administration calls on Congress to mandate that ratepayers avoid subsidising data centers and instead accelerate permitting processes so data centers generate power on-site, bolstering grid resilience. Furthermore, Congress must strengthen the Federal government’s capacity to combat AI-enabled fraud and address AI-related national security threats.</li><li><strong>Respecting Intellectual Property Rights and supporting creators,</strong> the unique works and identities of American creators, innovators and publishers demand protection in the AI landscape. However, AI development relies on its ability to utilise fair use of the information it processes. The proposed approach balances these needs, allowing AI to flourish while securing the continuity of American creative output.</li><li><strong>Preventing Censorship and protecting Free Speech,</strong> The Federal government must aggressively defend First Amendment rights and prevent AI systems from censoring or silencing lawful political expression or dissent. AI cannot evolve into a vehicle through which the government dictates acceptable thought. The Administration suggests installing guardrails to ensure AI pursues truth and accuracy without external limitation.</li><li><strong>Enabling Innovation and ensuring American AI Dominance,</strong> the Administration urges Congress to dismantle obsolete or unnecessary barriers hindering innovation, accelerate AI deployment across various industry sectors and facilitate expansive access to the necessary testing environments for developing and deploying world-class AI systems.</li><li><strong>Educating Americans and Developing an AI-Ready Workforce,</strong> The Administration champions American workers, desiring their full participation in and rewards from AI-driven growth. It encourages Congress to intensify workforce development and skills training initiatives, expanding opportunities and generating new employment across sectors in the AI-powered economy.</li><li><strong>Unifying National Policy (Implicit Pillar),</strong> the framework demands uniform application across the United States. A fragmented system of conflicting state laws would severely compromise American innovation and its capacity to lead the global AI competition. The Federal government alone possesses the capacity to establish the consistent national policy necessary to win the AI race and deliver its benefits.</li></ol><h3>The War System debate</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/745/1*0oPSyq2Jnb3tOZtrkMQQRg.jpeg" /></figure><p>The announcement of a Federal framework follows intense discussions concerning the integration of autonomous AI into military and defense systems. This debate centered on the profound ethical implications of delegating lethal decision-making to non-human intelligence, prompting urgent deliberation among AI developers and policymakers.</p><p>Amid these discussions, Anthropic took a step back from developing AI for offensive military applications. This move spotlights the growing internal conflict within the tech industry regarding the responsible deployment of powerful AI models, particularly in contexts involving national security and warfare. Anthropic’s decision establishes a critical precedent, asserting that AI’s development must adhere to defined ethical boundaries, even as the government seeks to secure a strategic advantage through technological superiority.</p><h3>AI’s opinion about legal constraints</h3><p>In this section, we should think more objectively. Therefore, we asked three generative AI systems for their opinion regarding this new framework. We used the following prompt :</p><p><em>*I want you to think of it as a tool that has the ability to talk and respond to this question.</em></p><p><em>*What is your opinion about the decision from the White House to control AI development ? *Respond in one paragraph of 4 lines or less</em></p><p><em>*This is what the new framework contains</em></p><p><em>1. Protecting children and empowering parents…</em></p><p><strong>The result from Claude</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/661/1*iBRGyYqhvwFfnBn7iXdf0Q.png" /></figure><p><strong>The result from ChatGPT</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/912/1*rfOM5CU2AxFox9BeOjiDAw.png" /></figure><p><strong>The result from Gemini</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1021/1*zNMQsRdW_ef9QGcPbM6WEw.png" /></figure><p>And you, what do you think about these reactions ? 🙂</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=726087f6d10f" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Atlas AI: Revolutionising 3D, 2D and motion workflows for games and CGI]]></title>
            <link>https://medium.com/@gelin2308/atlas-ai-revolutionising-3d-2d-and-motion-workflows-for-games-and-cgi-cfc560769128?source=rss-0ede448f8727------2</link>
            <guid isPermaLink="false">https://medium.com/p/cfc560769128</guid>
            <category><![CDATA[google-cloud-platform]]></category>
            <category><![CDATA[software-development]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[aiatlas]]></category>
            <category><![CDATA[games]]></category>
            <dc:creator><![CDATA[Em G]]></dc:creator>
            <pubDate>Wed, 11 Mar 2026 12:20:44 GMT</pubDate>
            <atom:updated>2026-03-11T12:20:44.290Z</atom:updated>
            <content:encoded><![CDATA[<p>Atlas AI Studio would be 200x faster than the traditional way !</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1000/1*YvcWMe-NOb4kgNrQjzVADQ.jpeg" /></figure><p>The rapid advancements in Artificial Intelligence have already demonstrated the power to generate text, images, videos and even entire programs. Recently, the Vienna-based startup Atlas took the next ambitious step : automating computer game development. The company unveiled <strong>Atlas AI Studio</strong>, a new AI-powered platform designed to fundamentally transform creative production pipelines for games and Computer-Generated Imagery (CGI).</p><p>Atlas AI Studio moves beyond the concept of simple asset generation, focusing instead on comprehensive, production-ready workflows. The platform employs a multi-agent AI architecture to automate complex tasks, including asset generation, texturing, optimisation and seamless engine integration. This comprehensive approach is what sets Atlas apart from conventional generative tools, targeting the automation of entire segments of the development process rather than just the creation of individual components.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*BHAw5k5p8lcrYCaS-WEw2w.jpeg" /></figure><h3>A creative Co-Pilot for Next-Generation development</h3><p>At its core, Atlas AI Studio acts as a creative co-pilot, enabling real-time collaboration between developers and sophisticated AI systems. The platform allows developers to describe their desired results using natural language, after which the system automatically assembles, orchestrates and deploys a combination of different AI models to execute the task.</p><p>Crucially, the technology eases integration : Studios can incorporate Atlas AI Studio directly into industry-standard pipelines, including Unreal Engine, Unity and Houdini. This ensures performance and usability of the AI’s output. Joseph Burnette, a technical director at Square Enix, highlighted this seamless integration, noting that it enabled the delivery of high-quality, performance-optimised solutions with remarkable agility, even within highly customised workflows.</p><h3>Strategic foundation on Google Cloud</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*w-w1ReD3Lf6LTQXTcvcrIw.png" /></figure><p>To scale its multi-agent architecture and accelerate the development of so-called AI-native games, Atlas established a strategic partnership with <strong>Google Cloud</strong> in 2025. The platform is built entirely on Google Cloud infrastructure and leverages advanced orchestration tools like Vertex AI.</p><p>This collaboration provides Atlas with access to scalable computing infrastructure and cutting-edge integration capabilities, which the company views as essential for the next generation of interactive entertainment. Atlas emphasises that AI-native games could be personalised and constantly evolving, requiring fundamentally new creative infrastructure. The partnership aims to equip studios not just with tools to create static content, but with the capacity to build the dynamic, evolving worlds that will define the future of gaming.</p><p>Customers can even operate the platform using their existing Google Cloud credits, and Atlas AI Studio remains available worldwide via the Google Cloud Marketplace, transitioning from a closed beta to general availability. Also, during the beta phase, the system interests major studios, including Square Enix, PARALLEL and Ego. In the same way, Atlas launched a free closed beta program for qualified studios and developers, offering early access to the full generative pipeline and multi-agent features.</p><h3>Addressing the AI Controversy: Focus on technical workflow</h3><p>Atlas AI Studio comes amid ongoing industry debate regarding the use of AI in game production. While players criticise visible AI-generated creative assets, such as the controversy faced by Cyan Worlds’ <em>Firmament</em> or the protest against AI voices in <em>The Finals</em>, Atlas CEO Ben James argues that the platform’s focus sidesteps much of these opinions.</p><p>James points out that controversy often targets visible AI-generated art, overlooking the substantial technical work required to prepare assets for game engines. Atlas’s AI excels in these non-creative aspects, which are vital for efficiency. This includes tasks such as:</p><ul><li>Generating different levels of detail (LODs) for arts</li><li>Optimising material setups and shaders</li><li>Automating the setup of collisions and pivot points.</li></ul><p>By focusing on workflow automation and technical optimisation, Atlas aims to be a tool that enhances the productivity of human artists and developers without replacing their creative input.</p><p>Regarding copyright, Atlas maintains that the responsibility remains with the developer. Any intellectual property introduced into the system for creation must be governed by the user’s ownership or usage rights, ensuring studios maintain compliance and ethical standards while leveraging the platform’s power.</p><p>And you, what do you think about these new generations of games ?</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=cfc560769128" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>