<?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 Dr. Thomas Jreige on Medium]]></title>
        <description><![CDATA[Stories by Dr. Thomas Jreige on Medium]]></description>
        <link>https://medium.com/@thomasjreige?source=rss-0c9ba822d379------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*kecg3ttzb0HX2WL54ljpWw.jpeg</url>
            <title>Stories by Dr. Thomas Jreige on Medium</title>
            <link>https://medium.com/@thomasjreige?source=rss-0c9ba822d379------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sun, 22 Nov 2026 03:31:15 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@thomasjreige/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[Preparedness advice needs its context.]]></title>
            <link>https://thomasjreige.medium.com/preparedness-advice-needs-its-context-f51d00b9a5c8?source=rss-0c9ba822d379------2</link>
            <guid isPermaLink="false">https://medium.com/p/f51d00b9a5c8</guid>
            <category><![CDATA[psychology]]></category>
            <category><![CDATA[resilience]]></category>
            <category><![CDATA[business-continuity]]></category>
            <category><![CDATA[disaster-recovery]]></category>
            <category><![CDATA[preparedness]]></category>
            <dc:creator><![CDATA[Dr. Thomas Jreige]]></dc:creator>
            <pubDate>Thu, 05 Nov 2026 11:19:12 GMT</pubDate>
            <atom:updated>2026-11-05T11:19:12.721Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="Three glass water carafes filled to different levels on a pale limestone shelf beside a small brass torch, lit by soft morning light from a tall window." src="https://cdn-images-1.medium.com/max/1024/0*iDz_NRm-OalZqEQ6" /><figcaption>Three days, a week, ten days: each figure rests on its own assumptions.</figcaption></figure><blockquote>How a board can read a public warning for what its sources establish.</blockquote><p>A director opens a message from a colleague late in the evening. It holds a map. More than thirty countries are marked on it, beneath a headline about households stocking up. The note above it says only: have you seen this?</p><p>Most directors will know the moment. A capable person has been handed an image that carries a great deal of urgency and very little else.</p><p>The advice on the map is ordinary and public. On 22 October the UK government posted its Be Ready guidance, which asks households to keep three days of water and food. The European Commission set a 72-hour benchmark in its Preparedness Union Strategy on 26 March 2025. Other governments publish their own figures, and one widely circulated compilation reports about a week in Sweden and Norway and up to ten days in Germany.</p><p>Those documents were written in different years, for different countries and for different purposes. Japan’s request for three days of supplies, and a week if possible, sits alongside named seismic scenarios such as a Nankai Trough megaquake. The UK guidance refers to storms, wildfires and the possibility of cyber disruption to essential services. A figure in days is the last link in a chain of assumptions, and a map shows only that last link.</p><p>The sources often carry their own context. The UK Defence Secretary told the BBC that the advice resembled guidance from the previous government and was not triggered by any particular piece of intelligence. The Security Minister described the campaign as bringing the UK into line with many of its European peers. A retail industry representative told a parliamentary committee that panic buying is quite easy to generate. Guidance that names its scenario leaves less room for that.</p><p>Scope matters as well. On 30 September the US Defense Secretary said that every major military installation in America would generate its own power, independent of the civilian grid. That is continuity planning for military installations. The published reporting on it does not address households, yet some coverage places it beside household guidance. A household kit, a disaster recovery plan and the operational continuity of a base have different requirements, and each answers a different question.</p><p>A pharmacist’s label offers a useful lens. It states what the medicine is for, how much to take, for how long and when to review. A public warning can be read in the same way, with four questions.</p><ul><li>What scenario is the advice written for, and does the issuer name it?</li><li>When was the advice first issued, and what has changed since?</li><li>Why this quantity and this duration, and who set them?</li><li>What does the issuer say it does not yet know?</li></ul><p>A board that asks these four questions can respond in proportion. The range of responses is wide. A board may review staff and travel arrangements, brief its people using the issuer’s own wording, check supplier continuity, or decide that no change is warranted and record the reasons. Each of these is a decision resting on evidence the board holds.</p><p>The questions also work in the other direction. When a board asks its own people to prepare, the people reading that message will ask the same four.</p><p>Urgency is most persuasive when the evidence behind it can be followed.</p><p>We work with boards and principals on decisions of this kind, where crisis preparation, decision readiness and communication meet. The practice reads the primary document, dates it and sets it beside the decision the board has to make.</p><p>When the decision needs to withstand scrutiny, begin with a conversation. hello@kaisira.com</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f51d00b9a5c8" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[When booking details make a scam believable.]]></title>
            <link>https://thomasjreige.medium.com/when-booking-details-make-a-scam-believable-e4c240d0444f?source=rss-0c9ba822d379------2</link>
            <guid isPermaLink="false">https://medium.com/p/e4c240d0444f</guid>
            <category><![CDATA[psychology]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[information-security]]></category>
            <category><![CDATA[cybersecurity]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Dr. Thomas Jreige]]></dc:creator>
            <pubDate>Thu, 29 Oct 2026 10:32:53 GMT</pubDate>
            <atom:updated>2026-10-29T10:32:53.497Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*M64ymMqfi-2NA16sMyTKug.png" /><figcaption>Familiar booking details can make a fraudulent payment request persuasive.</figcaption></figure><p>The RMS incident shows how customer information can lend credibility to a fraudulent payment request.</p><p>The message arrives before a holiday. It names the accommodation, gives the correct dates and refers to a booking the guest recognises. It asks for payment details and warns that the reservation could be cancelled.</p><p>There is enough familiar information to make someone pause. There may also be enough pressure to make them act.</p><p>For a traveller who has already organised leave, transport and accommodation, the prospect of losing a booking carries weight. A fraudulent request can draw its credibility from the arrangements the traveller has already made.</p><h3>What RMS has disclosed.</h3><p>In its statement dated 28 October 2026, RMS said it identified unauthorised access to a limited part of its environment on 26 October. Its current findings involve a subset of guest information, including names, contact details and booking details.</p><p>RMS says credit card and payment details were not involved. It has also warned of fraudulent email, SMS and WhatsApp messages that reference genuine bookings and seek payment information. The investigation remains ongoing.</p><p><a href="https://support.rmscloud.com/hc/en-gb/articles/18149918999823-Security-Incident-Alert-WhatsApp-Phishing-Scam">Source: RMS security incident statement</a></p><h3>Familiar details can lower suspicion.</h3><p>A booking record has value beyond the administration of a stay.</p><p>It can establish a connection between a person, a property and a particular time. In the wrong hands, that context may help someone impersonate a business the guest expects to hear from.</p><p>The guest may already be anticipating an arrival instruction, a balance reminder or a confirmation request. A convincing message can fit into that ordinary exchange.</p><p>This is why the accuracy of a message’s contents cannot establish the identity of its sender. A correct booking reference is information someone possesses. It does not, by itself, give that person authority to request a payment.</p><h3>The business still faces the guest.</h3><p>Accommodation operators rely on external platforms to manage reservations and communicate with customers. Those arrangements place parts of the customer relationship across several organisations.</p><p>When an incident occurs, guests will usually turn to the property they booked.</p><p>The reservations team needs to answer whether a message is genuine. Management needs to establish which records may be involved. The platform provider needs to supply timely evidence. Someone needs authority to approve a warning while the investigation is still developing.</p><p>These responsibilities deserve attention before an incident.</p><p>A supplier review should examine what information is held, how access is controlled and how long records are retained. It should also establish what the business can obtain when something goes wrong.</p><p>Can the supplier identify affected records? Who receives an urgent notification? Can the operator contact guests promptly? Who decides what can be said, and what remains uncertain?</p><p>The answers need named owners and a process that staff can use under pressure.</p><h3>Give customers a reliable way to check.</h3><p>A warning is more useful when it tells the guest exactly how to verify a request.</p><p>Businesses should make their normal payment arrangements clear and provide a dependable route for checking an unexpected message. Front-desk and reservations staff should be able to explain that route consistently.</p><p>Communications also need care. An urgent warning that introduces an unfamiliar payment or verification link may add to the confusion. The customer needs a simple way to reach a known contact and establish what action, if any, is required.</p><p>For travellers, RMS advises contacting the accommodation provider through confirmed official details before acting on a suspicious request. Avoid links in the message. If payment details have already been supplied, contact the financial institution immediately.</p><p><a href="https://support.rmscloud.com/hc/en-gb/articles/18149918999823-Security-Incident-Alert-WhatsApp-Phishing-Scam">Source: RMS guidance for travellers</a></p><h3>Customer information carries responsibility.</h3><p>The RMS incident gives businesses a reason to examine how their suppliers hold and use customer records, and how quickly they could respond to suspected misuse.</p><p>The discussion should reach beyond the technology team. Reservations, operations, management and communications each hold part of the response.</p><p>A useful question for leadership is straightforward: if someone used accurate customer information to impersonate this business tomorrow, how would the customer establish whom to trust?</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e4c240d0444f" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Advice changes faster than itineraries.]]></title>
            <link>https://thomasjreige.medium.com/advice-changes-faster-than-itineraries-bdbeab297aeb?source=rss-0c9ba822d379------2</link>
            <guid isPermaLink="false">https://medium.com/p/bdbeab297aeb</guid>
            <category><![CDATA[geopolitical-risk]]></category>
            <category><![CDATA[saudi-arabia]]></category>
            <category><![CDATA[board-decision]]></category>
            <category><![CDATA[duty-of-care]]></category>
            <category><![CDATA[principal-travel]]></category>
            <dc:creator><![CDATA[Dr. Thomas Jreige]]></dc:creator>
            <pubDate>Tue, 27 Oct 2026 06:58:48 GMT</pubDate>
            <atom:updated>2026-10-27T06:58:48.473Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*QNbyeSnwMfbUHzjGiJJSlg.png" /><figcaption>An empty travertine corridor in an airport at dawn, with soft light across a limestone floor and a single walnut bench.</figcaption></figure><p>How principal travel decisions are made when a transit hub becomes a target.</p><p>A journey through Riyadh was approved weeks ago. The hotel is confirmed, the meetings are set and the car is arranged. Everyone involved did the right things at the time they were done.</p><p>On Friday 16 October, Australia’s Smartraveller raised its advice for Saudi Arabia to “do not travel”, citing a volatile security situation. The advice states: “You should postpone any planned travel to Saudi Arabia, including travel for religious purposes.” The itinerary on file still reflects the risk as it stood on the day the trip was approved.</p><p>The reporting behind the change is specific. The Canberra Times, drawing on Australian Associated Press, reported that a spokesman for the Saudi-led military coalition counted more than 20 Houthi strikes on King Khalid International Airport in Riyadh during October. A separate attack on 10 October, reported by Reuters, killed 13 people and wounded more than 300, and was the deadliest in the kingdom in more than a decade. The Houthis have said they regard the kingdom’s airports and airspace as legitimate targets, with the exception of Mecca and Medina.</p><p>An airport that is a target changes the nature of a journey. A transit point often receives the least planning. It sits between the decision to travel and the meeting itself, and it is frequently chosen for convenience. When the transit point is the place under attack, the convenient routing becomes the most exposed part of the trip.</p><p>The official advice shows how quickly the practical questions multiply. It suggests avoiding the capital’s airport, taking extreme caution at other airports, minimising time in Saudi airports and leaving for the airport only once flights are reconfirmed. It also points to private car, train or bus travel, including to hubs in the United Arab Emirates and Qatar, where that is safe. Travellers are asked to be ready to shelter in place at short notice. Each of these is a decision that someone in the organisation needs to own.</p><p>Senior government minister Chris Bowen said the advice is “always appropriately conservative” and that Australians in Saudi Arabia can approach the embassy and consulate. A conservative government posture sets a baseline. An organisation’s own appetite for risk may sit above or below it, and a board benefits from knowing which before a traveller is in the air.</p><p>Three decisions sit beneath a change of this kind. The first is whether the journey proceeds, and who is entitled to say so. The second is how the routing changes if it does, including where the traveller waits and for how long. The third is who has the authority to alter or end the plan once the traveller is in motion, and how that person would learn that it is needed.</p><p>Travel approvals are often made once, at booking. Official advice may change several times between booking and departure. An organisation that treats approval as a standing condition, reviewed against current advice, is better placed to show why a journey proceeded. That record may matter later to insurers, to the board and to the traveller’s family.</p><p>For boards and principals, travel in a region with active conflict is a risk decision, and one that a board may be asked to explain.</p><p>We advises on principal and personnel travel as part of its geopolitical and strategic advisory work. That includes country and corridor risk assessment, in-country exposure analysis and the development of travel policy that fits the risk profile of the people and assets involved. The practice works alongside the security, legal and travel advisers already in place and examines the space between them.</p><p>Three questions are worth asking before any journey in the region. What would the organisation do if the airport closed while the traveller was in the terminal? Who would make that call? How quickly would that person know?</p><p>When the decision needs to withstand scrutiny, begin with a conversation. <a href="/service/mailto:hello@shimazakisentinel.com">hello@shimazakisentinel.com</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=bdbeab297aeb" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Board Has an AI Problem.
Or Does It?]]></title>
            <link>https://thomasjreige.medium.com/the-board-has-an-ai-problem-or-does-it-3df17c348079?source=rss-0c9ba822d379------2</link>
            <guid isPermaLink="false">https://medium.com/p/3df17c348079</guid>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[cybersecurity]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[information-technology]]></category>
            <dc:creator><![CDATA[Dr. Thomas Jreige]]></dc:creator>
            <pubDate>Wed, 21 Oct 2026 08:53:36 GMT</pubDate>
            <atom:updated>2026-10-21T08:53:36.426Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="A limestone aperture opens from a quiet walnut-lined room towards a distant landscape." src="https://cdn-images-1.medium.com/max/1024/0*WUGAOCWMU2U2_sBK" /></figure><p><em>When a new technology commands the boardroom, the more important question may be what is receiving less attention.</em></p><p>How much time has your board spent discussing artificial intelligence in the past twelve months?</p><p>Now consider how much time it has spent examining the integrity, ownership, sensitivity and commercial consequence of the information on which the organisation depends.</p><p>The answers may be revealing.</p><p>Artificial intelligence has become a substantial governance conversation. Policies are being written. Committees are reconsidering their responsibilities. Directors are attending briefings, management teams are preparing strategies, and advisers are offering increasingly elaborate frameworks.</p><p>Much of this attention is understandable. The technology is advancing quickly, its applications are expanding, and some consequences are difficult to anticipate. Yet a more fundamental question deserves consideration.</p><blockquote><strong>Does the board understand the information on which the organisation depends any better today than it did before artificial intelligence became a standing agenda item?</strong></blockquote><p>For some organisations, the answer may be uncomfortable.</p><h3><strong>The information was already there</strong></h3><p>Long before artificial intelligence entered the boardroom, organisations were accumulating information of considerable value: customer records, commercial agreements, intellectual property, financial information, operational intelligence, strategic plans and sensitive correspondence. The knowledge and experience embedded within the organisation mattered too.</p><p>That information was already being collected, stored, processed, shared and relied upon. It already carried obligations. It already presented risks.</p><p>In many organisations, the understanding of where it resides, who controls it, how reliable it is and what might follow from its misuse remains incomplete.</p><p>Artificial intelligence has brought renewed attention to these questions, often under the banner of an entirely new governance challenge. But the underlying information has not suddenly become valuable because an AI system can process it.</p><p>Its value existed before the technology arrived. So did much of the exposure.</p><figure><img alt="Walnut archive shelves and a limestone reading ledge hold a small collection of unmarked records." src="https://cdn-images-1.medium.com/max/1024/0*zsn2HPjNsbX7o7LF" /></figure><h3><strong>A new risk or an old weakness?</strong></h3><p>Artificial intelligence can introduce new exposures and change familiar ones. A system may produce a convincing but incorrect conclusion. It may infer sensitive information from material that appears harmless in isolation. It may create dependencies on external providers whose operations are poorly understood. Systems permitted to act autonomously may initiate actions before a person has considered their consequences.</p><p>These are legitimate matters for board consideration. They also require a distinction between exposures introduced by a particular use of AI and weaknesses that its use makes more visible or consequential.</p><p>An organisation that cannot identify its sensitive information has an information governance problem. One that does not understand who can access critical records has an access and accountability problem. One that relies on inaccurate or incomplete information for material decisions has an integrity and decision-quality problem.</p><p>Introducing artificial intelligence may increase the speed, scale and reach of the consequences. It does not necessarily change the underlying weakness.</p><blockquote><strong>Boards should distinguish a genuinely new exposure from an old weakness wearing a new name.</strong></blockquote><p>The distinction is not an argument for doing less. It is a basis for directing effort where it can improve assurance. Without it, organisations risk building additional layers of governance around problems they have yet to understand.</p><figure><img alt="Soft daylight reveals an existing joint between an old limestone wall and a bronze-framed glazed passage." src="https://cdn-images-1.medium.com/max/1024/0*krk3pvjH9i82xigt" /></figure><h3><strong>The cost of attention</strong></h3><p>Board attention is finite. Capital allocation, financial performance, operational resilience, regulatory obligations, leadership and strategic exposure compete for the same limited time.</p><p>Artificial intelligence has become a formidable competitor. Strategies require endorsement. Policies require review. Maturity assessments and new oversight arrangements demand consideration. Each may have merit. Collectively, they can consume substantial board and management capacity.</p><p>The question is whether that investment is proportionate to the organisation’s actual exposure.</p><p>A business using AI to assist with routine administration faces different consequences from one relying on automated systems to inform credit decisions, influence clinical outcomes, manage critical infrastructure or execute financial transactions. Even an administrative use can be consequential if it involves sensitive information or authority to act.</p><p>The technology may share a name. The consequences do not.</p><p>A board should understand where AI is used, what it is permitted to do, what information it depends upon and what might happen if it fails. The depth of examination should reflect the materiality of those consequences, the uncertainty involved and the organisation’s ability to intervene.</p><p><strong>The consequence determines the attention. The technology does not.</strong></p><figure><img alt="A limestone water channel passes beneath a single travertine crossing in a quiet courtyard." src="https://cdn-images-1.medium.com/max/1024/0*__f5K58_cj8DAZAt" /></figure><h3><strong>When governance becomes an activity</strong></h3><p>There is comfort in the visible machinery of governance. A policy can be approved. A committee can be established. A framework can be adopted. A dashboard can be presented.</p><p>These activities demonstrate that a subject has received attention. Whether they provide assurance is another question.</p><p>A board may receive a comprehensive AI governance presentation while remaining uncertain about the reliability of the information entering those systems. It may approve an acceptable-use policy without understanding which external providers receive commercially sensitive material. It may receive a maturity score without knowing whether a material decision could be explained, challenged or reversed.</p><p>Another governance document does not, by itself, resolve these shortcomings.</p><p>Effective oversight requires management to explain the exposure in terms directors can evaluate:</p><p>What information is involved, and what decisions depend upon it?</p><p>What could go wrong, and what would the consequences be?</p><p>Who has authority to act, and who remains accountable?</p><p>What controls are relied upon, and what evidence demonstrates that they work?</p><p>The evidence matters. An approved policy establishes an expectation. Assurance requires evidence that the expectation is being met in the processes on which the organisation relies.</p><p>These are familiar questions. Their familiarity does not diminish their importance.</p><h3><strong>Start with the asset</strong></h3><p>For many organisations, a more useful starting point is the information itself.</p><p>Establish what matters and why: its commercial, operational, legal and strategic significance. Determine who is responsible for it, where it moves, who can access it and how its integrity is maintained.</p><p>Then examine what changes when it enters an AI-enabled process.</p><p>Is information exposed to another party? Can it be retained, reused or combined to infer more than the organisation intended?</p><p>Can outputs be verified? Could an inaccurate output influence a consequential decision?</p><p>Can a person intervene before an action becomes irreversible?</p><p>What dependencies are introduced, and what happens if they fail?</p><p>This approach does not eliminate the need for AI-specific controls. It establishes a basis for deciding which controls are necessary, who should oversee them and how much assurance they require.</p><p>It also returns the discussion to the organisation’s circumstances. The same technology can represent a modest exposure in one setting and a significant strategic or operational risk in another. The board needs a clear explanation of the difference.</p><figure><img alt="A closed leather notebook and a bronze straightedge rest on a stone reading ledge beside a deep window." src="https://cdn-images-1.medium.com/max/1024/0*4YQCuNiOsn1M7QfD" /></figure><h3><strong>Judgement remains with the board</strong></h3><p>Artificial intelligence can analyse information, identify patterns, generate recommendations and perform tasks that previously required human intervention.</p><p>A recommendation remains only as useful as the evidence, assumptions and limitations on which it rests. A decision does not become defensible simply because a sophisticated system contributed to it.</p><p>Boards need assurance that material decisions are made under appropriate authority and that their consequences are understood. Management needs clear responsibility for the processes through which those decisions are made and implemented.</p><p>That requires more than familiarity with the technology. It requires an understanding of what can reasonably be relied upon.</p><p>Where information is incomplete, the limitation matters. Where conclusions cannot be adequately tested, the uncertainty matters. Where an automated action can create consequences beyond the organisation’s ability to intervene, the allocation of authority matters.</p><p>These are questions of governance, risk and judgement. They deserve attention because of their consequences.</p><h3><strong>Perhaps the board has a different problem</strong></h3><p>Artificial intelligence deserves serious consideration. In some organisations, it will require substantial changes to oversight, operating models and the allocation of responsibility. In others, the appropriate response may be more modest.</p><p>The challenge is knowing which circumstances apply.</p><p>That determination should emerge from a disciplined understanding of the organisation’s assets, exposures, dependencies and decisions. It should include an examination of the governance arrangements already in place and whether there is evidence that they address the risks identified.</p><p>There is considerable experience to draw upon. Boards have long overseen information, operational systems, external dependencies and consequential decisions. The arrival of AI has made some consequences of misunderstanding them potentially greater.</p><p>Independent scrutiny can feel uncomfortable once time and capital have been invested. That is also when testing the assumptions behind an approach may be most valuable.</p><p>Before commissioning another framework or adding another standing agenda item, a board might reasonably ask what is genuinely at risk, what remains uncertain and whether the attention devoted to AI is improving its ability to make sound decisions.</p><blockquote><strong>Is the board governing the organisation’s actual exposure, or simply the subject that currently commands the most attention?</strong></blockquote><p>That distinction may be worth more than another AI presentation.</p><p><em>Shimazaki Sentinel | Independent judgement for decisions that must withstand scrutiny.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3df17c348079" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Adult in the Room Is Also the Salesman]]></title>
            <link>https://thomasjreige.medium.com/the-adult-in-the-room-is-also-the-salesman-bf97e07a9393?source=rss-0c9ba822d379------2</link>
            <guid isPermaLink="false">https://medium.com/p/bf97e07a9393</guid>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[risk-management]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[psychology]]></category>
            <category><![CDATA[cybersecurity]]></category>
            <dc:creator><![CDATA[Dr. Thomas Jreige]]></dc:creator>
            <pubDate>Fri, 18 Sep 2026 06:49:24 GMT</pubDate>
            <atom:updated>2026-09-18T06:49:24.586Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*occS8Oclfib_fl28ruMKNw.png" /><figcaption>The Adult in the Room Is Also the Salesman</figcaption></figure><p><em>When the people warning you about a technology are the same people selling it, the warning carries information about their position alongside whatever it says about the technology.</em></p><p>In the second week of September 2026, something rare happened in the artificial intelligence industry. People who agree on almost nothing agreed on something. Dario Amodei of Anthropic published an essay called “We Must Pace the Frontier,” arguing that the speed of AI development had become dangerous and needed deliberate slowing. Within about forty-eight hours Sam Altman of OpenAI and Elon Musk of xAI, competitors who spend most of the year describing each other in unflattering terms, had lined up behind it. For an industry built on the promise that it will change everything, the striking new consensus was that everything might be changing too fast.</p><p>Michael Burry was unpersuaded. The investor is best known for shorting the mortgage market before it collapsed in 2008, and for being played by Christian Bale in the film about it. Posting on X on 14 September, through the account he now runs under the name Cassandra Unchained, he offered a reading of the sudden agreement that was considerably less flattering to the people inside it. The safety warnings, he suggested, were self-serving. His argument ran along four lines. Large language models are not the general intelligence the marketing implies, so there is nothing there to meaningfully slow down. Any slowing of development protects the companies already in front. An initial public offering needs a story, and “we have built something so capable it frightens us” is an unusually good one. And the fear may be doing quieter work, covering for growth that is softening on its own while those public offerings get pushed further out.</p><p>I am not going to referee whether Burry has read their minds correctly. Whether he is right about the motives of Amodei, Altman and Musk is close to unknowable, and it is also close to beside the point. What he has done, whether he intended to or not, is put a very old question back on the table. Intelligence analysts have a phrase for it. Cui bono. Who benefits.</p><p>When a claim reaches you, the identity of the source is part of the evidence. A warning about the dangers of a product, issued by the maker of that product, in the season the maker is raising money, is not thereby false. Treating it as false because of who said it is its own mistake, the one logicians call the genetic fallacy. The origin of a claim does not settle whether it is true. A vendor can have every reason to talk his book and still be describing a real hazard. Sometimes the arsonist is the only person who has actually seen the fire.</p><p>So the source structure does not tell you whether the hazard is real. It tells you how much independent verification the claim still needs before you act on it. A risk signal from an interested party has value. What it lacks is corroboration. The failure that costs money and reputation is the quiet upgrade from “the vendor says this is dangerous” to “this is dangerous,” with the vendor’s interest deleted somewhere along the way.</p><p>Set motive aside completely, because the interesting part survives without it. The message “our technology is so powerful that it needs restraining” does specific work in the world regardless of the sincerity behind it. It casts the speaker as the responsible adult in a room full of children. It raises the imagined cost of entry for anyone smaller, because rules written for capability the incumbents already possess are affordable to the incumbents and punishing to a challenger. And it converts public anxiety into a market position. Fear, directed well, is a moat. None of this requires anyone to be lying. A sincerely frightened market leader and a purely cynical one write the same essay and profit from it the same way. That is what makes the behavioural signal so hard to read, and so easy to exploit.</p><p>Burry’s stronger material sits underneath the talk about motives, and it has the advantage of being checkable. For most of the past year he has argued that the large infrastructure buyers behind the AI build-out have been flattering their earnings by stretching the assumed useful life of hardware that runs on a two to three year cycle. Depreciate a graphics processor over six years instead of three and current profit looks better today and worse later, by an amount that analysts have placed in the region of a hundred and seventy-six billion dollars across the sector through the end of the decade. He has held short positions against Nvidia, Palantir and Tesla, and has warned of the possibility of a 1987-type fall in AI equities. That case can be tested against filings and disclosures. It does not depend on knowing what anyone was thinking. It is the epistemically clean half of his argument, and it is the half most worth a serious reader’s attention, because it points directly at where growth would show strain if it were beginning to slow.</p><p>The governance problem begins here. Boards are being asked, right now, to make decisions about adopting, funding and depending on frontier AI. A great deal of the risk intelligence available to them on the subject is produced by the companies selling the technology, or by the investors positioned against it. Both are interested parties. A director who discharges a fiduciary duty by leaning on a vendor’s safety narrative, or on a short seller’s alarm, has handed judgement to someone who holds a position in the outcome. Under the accountability regimes now operating in Australia and comparable jurisdictions, and against precedent that has held directors personally responsible for the reasoning behind their decisions, that is an uncomfortable place to stand.</p><p>The rule that follows is not complicated, though it is routinely ignored. When the only smoke detector in the building is sold by the company that also supplies the fire, you do not conclude there is no fire, and you do not take the salesman’s word for how big it is. You install a detector made by someone who does not care whether your building burns. Independent assurance exists for exactly the situations where every loud voice has something riding on what you come to believe. The present AI moment, in which rivals who agree on almost nothing have discovered a shared and lucrative concern for your safety, is close to a textbook example.</p><p>Burry may be trading his own book. The labs may be sincerely afraid of what they are building. Both of those things can be true at the same time, and neither of them tells a board what to do. What tells a board what to do is an assessment of the technology sourced from someone who profits from neither the hype nor the crash. That person is becoming harder to find, and that scarcity is the most useful thing this whole episode reveals.</p><p><em>Sources</em></p><ul><li><a href="https://www.thestreet.com/technology/michael-burry-slams-ai-slowdown-push">Michael Burry slams AI slowdown push — TheStreet</a></li><li><a href="https://www.mediaite.com/media/news/big-short-maven-michael-burry-argues-ai-companies-are-hyping-panic-to-cover-for-slowing-growth-in-the-industry/">Big Short maven argues AI companies are hyping panic to cover for slowing growth — Mediaite</a></li><li><a href="https://www.coindesk.com/tech/2026/09/12/anthropic-ceo-calls-for-ai-race-to-slow-down-musk-and-openai-s-altman-agrees">Anthropic CEO calls for the AI race to slow down; Musk and Altman agree — CoinDesk</a></li><li><a href="https://www.fool.com/investing/2026/09/12/michael-burry-is-staying-short-nvidia-palantir-and/">Michael Burry is staying short Nvidia, Palantir and Tesla, warning of a possible 1987-type fall — The Motley Fool</a></li><li><a href="https://www.cnbc.com/2025/11/11/big-short-investor-michael-burry-accuses-ai-hyperscalers-of-artificially-boosting-earnings.html">Big Short investor accuses AI hyperscalers of artificially boosting earnings — CNBC</a></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=bf97e07a9393" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The CrowdStrike Signature: What Two Years Should Have Taught the Board]]></title>
            <link>https://thomasjreige.medium.com/the-crowdstrike-signature-what-two-years-should-have-taught-the-board-454b9739b522?source=rss-0c9ba822d379------2</link>
            <guid isPermaLink="false">https://medium.com/p/454b9739b522</guid>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[information-security]]></category>
            <category><![CDATA[psychology]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Dr. Thomas Jreige]]></dc:creator>
            <pubDate>Tue, 01 Sep 2026 00:55:48 GMT</pubDate>
            <atom:updated>2026-09-01T00:55:48.091Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*7TqcFHI7Mp-0f4YqqoL26g.png" /><figcaption>The CrowdStrike Signature: What Two Years Should Have Taught the Board</figcaption></figure><p>At 04:09 UTC on 19 July 2024, CrowdStrike pushed a routine configuration update to its Falcon sensor. The change touched a single component, Channel File 291, the part of the sensor that screens named pipes on Windows. The file contained a logic error that produced an out-of-bounds memory read. On roughly 8.5 million Windows machines, the sensor was running in the kernel, so the fault did not degrade a service or slow a process. It halted the operating system. Machines dropped into bootloops and recovery mode across the world within the same few minutes, because the update carried no staged rollout and no option to defer it. Every device took it at once.</p><p>There was no attacker. CrowdStrike’s own chief executive confirmed within hours that this was not a cyber event. A trusted supplier shipped a flawed file through a trusted channel, and the design of that channel did the rest. Airlines grounded fleets and worked the gates on paper. Hospitals stood down elective lists. Payment, broadcast and logistics systems failed together, for the same reason, at the same moment.</p><p>Two years on, most reviews of that day still file it under cybersecurity. That is where the analysis goes wrong. What happened on 19 July was a governance event with a technical trigger, and the exposure it revealed is one that sits with the board. The useful work now is to understand precisely why it was so severe, because the shape of that severity tells a director exactly what to look for in their own organisation.</p><h3>The signature that made it catastrophic</h3><p>Plenty of software is installed almost everywhere and never brings the world down when it fails. Ubiquity on its own is survivable. What made this event different was the combination of three properties held in a single supplier at the same time.</p><p>The first is reach. Falcon ran across a vast share of enterprise Windows estates, so a single fault had an enormous population to land on. The second is privilege. The sensor operated at kernel level, with the authority to stop the machine, which is why a content error became a system halt instead of an application crash. The third is autonomy. The update installed itself on every machine simultaneously, with no staged deployment and no human approval standing between the vendor’s decision and the customer’s operations.</p><p>Reach, privilege and autonomy in one place is the risk signature worth naming. Any supplier that holds all three has become a systemic point of control inside your organisation, whether or not it appears that way on a procurement register. Most concentration analysis counts spend or counts logos. The more revealing question is which of your suppliers can reach everything, act with privilege, and change themselves without asking. Very few will meet all three tests. The ones that do are the ones that can stop you.</p><h3>Concentration turns independent risks into one correlated risk</h3><p>The deeper failure that day was collective. Organisations went down together, and their recovery assumptions went down with them.</p><p>Almost every continuity plan is built on an unstated assumption of independence. We assume our data centre and our recovery site will not fail for the same reason on the same morning. We assume that if we are disrupted, our incident response partner, our cloud platform and the third parties we depend on to recover are still standing. Shared dependence quietly dissolves that assumption. When the common supplier fails, the failures correlate. The responder you would call is down. The recovery environment is degraded. In Delta’s case, the systems that schedule crews failed alongside everything else, which is a large part of why the airline needed the better part of a week to recover while others were back in a day.</p><p>The insurance industry has modelled this for a century and calls it aggregation. A reinsurer will not write every house along the same fault line without knowing its total correlated exposure, because the loss that matters is the one where all the policies pay at once. Boards carry the same kind of exposure across their technology base and rarely measure it that way. Concentration risk is an aggregation problem, and it belongs on the risk register as one.</p><h3>Why the liability lands on the board</h3><p>The most common defence offered inside affected organisations was that the vendor was reputable and widely used. That answers a procurement question. It does not answer the board’s.</p><p>The contract makes the point plainly. Delta has claimed around 500 million dollars in losses from those days of disruption. CrowdStrike’s public position is that its contractual liability is capped in the single-digit millions. Whatever a court eventually decides, the structure is the lesson. The consequences of a supplier failure do not transfer to the supplier along with the purchase order. They stay on the customer’s balance sheet, which is to say they stay with the directors who accepted the exposure.</p><p>This is where the fiduciary reality bites. The protection that directors rely on for a decision that later goes wrong covers a judgement they actually made on an informed basis, in good faith, believing it to be in the interests of the company. It does not extend to an exposure the board never examined at all. A concentration risk that was never mapped sits outside that protection. It is an unexamined liability carried as though it were ordinary operations.</p><p>For APRA-regulated boards the point is now written into standard. CPS 230 came into force on 1 July 2025 and requires regulated entities to manage the risks of their material service providers and to keep their critical operations within defined tolerance levels through severe disruption. The regulator has, in effect, made concentration and continuity a board-level obligation with a name attached. For every other board, the duty of care and diligence already implies the same thing. CPS 230 simply removes the argument that anyone still thought it optional.</p><h3>What did we actually learn</h3><p>Across most of the market, less than the day deserved. There was a season of attention, a round of internal reviews, and a set of assurances that the matter was in hand. Then it moved on. A smaller number of organisations did the real work and are materially harder to stop today. The honest test for a director is not whether a review was completed. It is whether the board can answer four questions with evidence:</p><ul><li>We can identify which suppliers hold reach, privilege and the autonomy to change our systems without our approval, and we treat those as systemic points of control.</li><li>We know, for each of them, how long we can keep operating and by what means if they became unavailable without warning, and we have set a tolerance for that disruption.</li><li>We have tested that continuity against correlated failure, where the supplier and the partners we would rely on to recover are down at the same time.</li><li>We hold this exposure as a standing board matter with a named owner, and we have accepted or reduced it as a deliberate, recorded decision.</li></ul><p>A board that can speak to those has learned from CrowdStrike. One that cannot has been fortunate, and fortune is not a control.</p><h3>The next one begins with a machine</h3><p>There is a reason this matters more now than as a two-year retrospective. Look again at the signature. Reach and privilege are structural and slow to change. Autonomy is the property currently accelerating, as security and infrastructure tooling moves toward automated and increasingly agentic change across estates that have grown steadily more concentrated. The human approval step that might have caught Channel File 291 was already absent in 2024. In an operating model where systems initiate their own changes, that step is removed by design, and the interval in which a person could intervene before a fault propagates shrinks toward zero.</p><p>That is the strategic case for treating concentration as live governance today. The exposure has not eased in two years. The speed at which it can be triggered has increased, and the direction of the technology is adding autonomy to suppliers that already have reach and privilege. Boards that map that signature, set tolerances, and test for correlated failure are making a defensible decision on behalf of the enterprise. Directors who leave it with the technology function are holding a liability they have chosen not to look at, on a risk that keeps growing while it waits.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=454b9739b522" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Securing Whose Fuel? Reading the Fine Print in Australia’s Cleaner Fuels Plan]]></title>
            <link>https://thomasjreige.medium.com/securing-whose-fuel-reading-the-fine-print-in-australias-cleaner-fuels-plan-33dc0f941cb8?source=rss-0c9ba822d379------2</link>
            <guid isPermaLink="false">https://medium.com/p/33dc0f941cb8</guid>
            <category><![CDATA[psychology]]></category>
            <category><![CDATA[government]]></category>
            <category><![CDATA[management]]></category>
            <category><![CDATA[energy]]></category>
            <dc:creator><![CDATA[Dr. Thomas Jreige]]></dc:creator>
            <pubDate>Thu, 27 Aug 2026 02:08:23 GMT</pubDate>
            <atom:updated>2026-08-27T02:08:23.235Z</atom:updated>
            <content:encoded><![CDATA[<blockquote><em>The government has named a real vulnerability. The instrument it has chosen to fix it deserves closer reading than the executive summary invites.</em></blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*hS934pLjs3NYrQSu_YQOmg.png" /><figcaption>Securing Whose Fuel? Reading the Fine Print in Australia’s Cleaner Fuels Plan</figcaption></figure><p>Australia depends on other countries for the fuel that moves almost everything. More than 90 percent of the liquid fuels we use are either imported directly or refined here from imported crude. Two refineries remain after two decades of closures. In a detail that should stop any reader, around three quarters of the canola we grow is shipped offshore, much of it feeding biofuel industries in other nations, while we buy the finished fuel back. Anyone responsible for continuity, whether in a boardroom, a supply chain, or a critical infrastructure operator, should treat that picture as a genuine exposure rather than an abstract policy concern.</p><p>So when the Department of Climate Change, Energy, the Environment and Water released its consultation paper, <em>Securing Australia’s Cleaner Fuels Industry</em>, the starting diagnosis was sound. The country is exposed to global price shocks and supply disruption, it captures little of the value in its own feedstocks, and its liquid-fuel emissions sit at roughly 31 percent of the national total. Those are real problems and they are worth acting on. The question this piece examines is whether the mechanism the government has chosen does what its title promises.</p><h3>What the paper proposes</h3><p>The paper sets out a demand mechanism for low carbon liquid fuels, the family that includes renewable diesel, sustainable aviation fuel, ethanol, biodiesel and synthetic e-fuels. It would oblige fuel suppliers to put a rising share of these fuels into the Australian market.</p><p>The scheme runs in two phases. From 2029 to 2035, suppliers face a volumetric mandate, meaning a fixed volume of low carbon fuel they must supply each year. From 2035, the scheme transitions to a carbon-intensity standard, where compliance is measured by the emissions a fuel saves across its full lifecycle, expressed in grams of carbon dioxide per megajoule. That later design follows established schemes in California, Canada and the United Kingdom. Coverage starts with diesel and jet fuel and extends to petrol, reaching about 95 percent of the fuel market. Obligated suppliers meet the requirement by supplying the fuel, surrendering credits, or paying a buy-out price. A separate aviation requirement runs throughout to keep demand flowing to sustainable aviation fuel.</p><p>The timeline is deliberate. Submissions close on 15 September 2026, legislation is drafted through 2027, compliance begins in 2029, and the first formal review is not scheduled until 2032. The government is also considering building the scheme into the existing Fuel Security Act rather than passing a standalone law, with much of the detail left to regulation.</p><h3>What the paper gets right</h3><p>It is worth being clear that this is a more honest document than most consultation papers, and several choices deserve credit.</p><p>The two-phase logic is defensible. A new domestic industry needs a predictable demand signal before it can attract investment, and a volume target during the formation years gives producers something concrete to plan against. Moving to a carbon-intensity basis later, once the market can bear it, rewards the fuels that actually save the most emissions.</p><p>The risk table is unusually candid. The department rates the chance that the policy distorts the market, favouring particular suppliers or pathways, as high. It rates geopolitical instability as high and admits in plain language that the risk is hard to manage effectively. Public papers rarely concede that much.</p><p>The integrity thinking is sound. The paper insists on lifecycle carbon accounting, calls for chain-of-custody tracing, and notes the documented cases where virgin palm and seed oils have been passed off as used cooking oil to claim a cleaner rating. Recognising that fraud risk upfront is the mark of people who have studied how these schemes fail elsewhere.</p><p>It has also learned from overseas. It notes that California’s credit market ran into sustained oversupply, which pushed credit prices down and weakened the investment signal, forcing later reforms. Designing with that lesson in view is prudent.</p><h3>The distance between the promise and the mechanism</h3><p>The paper is titled and framed around fuel security and domestic industry. The figures used to sell it are large, roughly 10 billion dollars in annual GDP and more than 26,000 jobs. Read past the summary and two admissions sit uneasily against that framing.</p><p>First, the mechanism does not require the fuel to be made in Australia. The department states this directly in section 3.4.2. It then concedes in section 3.4.1 that imported low carbon fuel may be more cost-competitive than much Australian production and could supply a substantial share of demand. Place those two sentences side by side and the security story weakens. A supplier facing an obligation will buy the cheapest compliant fuel available. If that fuel is imported, the policy has swapped dependence on imported fossil diesel for dependence on imported biofuel. The refineries, the jobs and the regional feedstock processing that justify the policy are outcomes the scheme hopes for. It does not secure them.</p><p>Second, the economic case that sells the policy sits outside the economic analysis that tests it. The cost-benefit analysis quantified only emissions reduction and air quality. Fuel security, industry development and the headline dollar figures were assessed qualitatively and are absent from the benefit-cost ratios, as the paper states in section 2.7.1. At the same time the paper declines to estimate what the scheme will do to fuel prices, saying it has no reliable basis for the figure. A mandate is a cost carried by the people who buy diesel, jet fuel and petrol, which means freight operators, farmers, miners, airlines and households. The costs are real and unstated. The benefits used in the marketing are not in the ledger.</p><p>A third point stands out. The department’s own modelling found that the more flexible carbon-intensity approach delivered better value than the volumetric mandate. The policy leads with the volumetric mandate anyway. The stated reason is investment certainty, which is a fair argument on its own terms. It is also an accurate description of guaranteed demand for a domestic industry, funded by fuel users, delivered through a more expensive instrument than the one the modelling preferred. Supporting an industry through a fuel mandate keeps the cost out of the federal budget and places it on the bowser. That is a legitimate policy choice. It should be named for what it is so that the people paying for it understand the trade being made.</p><h3>We have run this experiment before</h3><p>Australia has tried fuel mandates. New South Wales legislated an ethanol requirement from 2007 and Queensland followed. Both spent years below their legislated volumes. Consumers avoided the blended product, retailers took exemptions, and the local fuel needed continuing government support to stay in the market. A briefing from that era described biofuels as not cost-competitive with conventional alternatives and likely to need substantial and ongoing support. The new paper lists those state mandates among the policies it has to reconcile with, so the precedent is well understood inside government.</p><p>The cost premium is not a temporary glitch that scale will erase. A good part of it is structural. Petroleum has already been stripped of oxygen by geology. Plant oils, residues and captured carbon dioxide carry oxygen that has to be removed with energy and hydrogen before the result behaves like diesel or kerosene. Recent analysis puts sustainable aviation fuel at around 2.8 times the price of conventional jet fuel, and power-to-liquid e-fuels near ten times. Growth does not automatically close the gap, because the cheapest feedstock, waste oils and fats, is limited, so expansion pushes production onto dearer pathways. For liquid fuels specifically, a durable premium is the honest expectation, and any policy built on the hope of near-term price parity is building on sand.</p><h3>The question the paper does not ask</h3><p>Electrification runs through the document as the cheapest way to cut emissions, and it appears repeatedly as the pathway that could undercut demand for the very fuels the scheme is promoting. E-fuels, which the paper counts as a future compliance option, are made from hydrogen produced with electricity. The emissions saved by all of this depend entirely on how that electricity is generated. The paper never examines the generation mix. It specifies hydrogen from renewable electricity in its own fuel table and goes no further.</p><p>This is a strategic gap rather than a technical footnote. A policy designed to reduce liquid-fuel emissions leans on electricity whose carbon intensity it does not interrogate, and on future e-fuels whose viability rests on abundant zero-carbon power that Australia has not yet built. A serious plan for cleaner fuels, and for the electricity that increasingly sits behind them, should state where firm low-emissions generation is coming from and weigh the full range of options for supplying it, from firmed renewables through to the sources currently ruled out of the national conversation. Leaving the generation question out of scope makes the fuel plan read as one half of a strategy whose other half is missing.</p><h3>What a more defensible version would look like</h3><p>None of this argues for standing still. The exposure is real and worth addressing. From a risk and governance standpoint, a stronger version of this policy would do a few things the current draft avoids.</p><p>It would state the cost and who carries it. A mechanism that adds cost across 95 percent of the fuel market owes fuel users an honest estimate and a plan for the sectors least able to absorb it, particularly agriculture, freight and aviation.</p><p>It would align the instrument with the objective. If the aim is fuel security through domestic production, the obligation should reward fuel made and processed here, with the trade-law and cost consequences worked through in the open. If domestic production cannot be favoured for trade reasons, the security language should be softened and the policy defended honestly on its emissions merits.</p><p>It would sequence the fuel plan with a generation plan. A commitment to e-fuels and electrification is an implied commitment to firm, low-emissions power, and that belongs in the same conversation rather than a separate one.</p><p>It would keep the flexibility its own modelling rewards and give the review dates real force. A first review in 2032, three years after a 2029 start and long after the design is locked, arrives late. Clear triggers tied to fuel-price impact, to the share of compliance met by imports, and to actual domestic production built would let the scheme correct before the bill grows large.</p><h3>The bottom line</h3><p>The consultation paper is more honest than most. It admits its uncertainties, rates its risks candidly, and genuinely invites challenge. Its weakness lies in the distance between how it is framed and what it does. A decision of this scale, imposing a long-term cost across the economy on behalf of benefits it has not fully counted, would struggle to pass the kind of scrutiny a competent board applies to its own major commitments. The questions a director would ask are precisely the ones the paper has deferred. What will it cost. Who pays. What do we actually secure. And what happens to the power supply the entire design rests on.</p><p><em>Shimazaki Sentinel provides independent risk, assurance and strategic advisory to boards, executives and critical infrastructure operators. This article reflects our reading of the government’s consultation paper and is offered as commentary rather than legal or financial advice.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=33dc0f941cb8" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[He Wasn’t Laughing at the Chef]]></title>
            <link>https://thomasjreige.medium.com/he-wasnt-laughing-at-the-chef-c8fc05a00bff?source=rss-0c9ba822d379------2</link>
            <guid isPermaLink="false">https://medium.com/p/c8fc05a00bff</guid>
            <category><![CDATA[psychology]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[children]]></category>
            <category><![CDATA[business]]></category>
            <category><![CDATA[mindset]]></category>
            <dc:creator><![CDATA[Dr. Thomas Jreige]]></dc:creator>
            <pubDate>Mon, 24 Aug 2026 02:07:15 GMT</pubDate>
            <atom:updated>2026-08-24T02:07:15.062Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*PtK7xeiq-uv8YIx8YLUeKA.png" /><figcaption>He Wasn’t Laughing at the Chef</figcaption></figure><p>My grandson stayed over this weekend, and it was, as it always is, a pleasure and never a dull moment. We did the basic things. We built puzzles. We kicked a football around the park until neither of us could feel our toes. On Sunday morning we made pancakes, which has become our thing, the fixed point the whole weekend orbits around. I had it marked, quietly, as the highlight before it had even happened.</p><p>Somewhere in the middle of Saturday we ended up on the sofa with a few YouTube videos running. One of them was an old Muppets clip, the Swedish chef with a microwave stuffed full of popcorn, conducting the whole kitchen like an orchestra as it went off. I have seen that clip more times than I can count across several decades. I did not expect it to do anything to me at all.</p><p>Here is a link to the video if you dare to laugh :-)</p><p><a href="https://www.youtube.com/watch?v=B7UmUX68KtE">https://www.youtube.com/watch?v=B7UmUX68KtE</a></p><p>It undid me completely. I lost my composure laughing, properly laughing, the kind that catches you by surprise and will not let go. My grandson looked at me for a second, and then he went too. We ended up crying with laughter together on that sofa, and by Monday my wife was still singing variations of the popcorn song around the house, and I am apparently never going to hear the end of it.</p><p>Here is the part that has stayed with me since. He did not laugh at the chef first. He laughed because I did.</p><p>I have thought about why that detail matters more than it should. The pancakes were planned, anticipated, savoured in advance the way a tradition gets savoured. Their pleasure had already been partly spent before Sunday arrived, because I had been looking forward to them since Friday. The Muppets clip carried none of that weight. Nobody had built it up. It simply happened, and because nothing had been asked of it in advance, it was free to land at full strength. That is a fairly well established feature of how surprise works on us. The unexpected version of a good thing tends to hit harder than the anticipated one, because there is no gap between what we hoped for and what we got. There is just the thing itself, arriving unannounced.</p><p>But that only explains the clip. It does not explain the sofa.</p><p>A four year old is not yet equipped to follow the specific joke of an unhinged Swedish chef losing control of a kitchen appliance. What he is fully equipped to follow, from a very early age, is a break in pattern. A grandfather who is normally composed, normally the one setting the tone of the room, suddenly gone, wheezing, unable to speak. That is the incongruity his mind actually caught. Not the chef’s chaos. Mine. The joke he responded to was the gap between how I usually behave and how I was behaving in that moment, and once he had that, my laughter became more interesting to him than anything happening on the screen.</p><p>There is a reason his laughter arrived on the back of mine rather than beside it. People catch laughter far more readily from someone they are close to than from a stranger, and it registers as more genuine when they do. A posed laugh from someone we love still reads as authentic in a way a stranger’s genuine laugh does not quite manage. What passed between us on that sofa was not two people responding independently to the same video. It was one person’s loss of control becoming the other person’s cue, and the whole exchange working precisely because there was nothing manufactured about it. He was not laughing because the moment was engineered to make him laugh. He was laughing because he could tell, instantly and without needing to be told, that I actually meant it.</p><p>That is the part I keep returning to, in a week that has otherwise been full of the usual business of assessing whether things are what they claim to be. So much of what reaches all of us now, adults and children both, has been shaped to simulate exactly this kind of spontaneous delight. Content is engineered for reaction the way a room is staged for a photograph, timed, tested, optimised for the laugh or the tear or the share. My grandson has years ahead of him in an environment that will get steadily better at manufacturing moments built to feel like the one we had by accident. What he caught on that sofa was the real thing, uncurated, unplanned, and entirely mine to give him. No algorithm produced it. No one was trying to make either of us feel anything. It simply happened, and because it was genuine, he trusted it instantly and completely, the way children do when something has not been dressed up for their benefit.</p><p>There is a kind of intelligence in that trust that we spend less time noticing than we should. Long before a child can explain why something feels honest rather than performed, they can tell the difference. The skill does not leave us as adults so much as it gets buried under years of learning to be polite about things that do not quite ring true. He had not learned that yet. He simply responded to what was real and ignored what was not, which is a more sophisticated piece of judgement than a four year old gets credit for.</p><p>I catch myself doing a version of this at work too, at Shimazaki Sentinel, where a good part of my week involves working out whether the composed version of something is the real one or just a well rehearsed one. Boards and executive teams do it too, without meaning to, and the moments that actually build trust in a room are rarely the ones anyone planned or polished. They are the moments someone drops the script for a second and means what they say. My grandson worked that out on a sofa before he could tie his own shoelaces. It took me a good deal longer, and I still find it useful to be reminded by a four year old.</p><p>My wife is still singing that song. I expect she will be for some time yet, and I have decided I do not mind. It has become part of the furniture of our house now, a small, ridiculous, permanent fixture, which is its own kind of evidence. The things that get repeated back to us, that earn a place in a family’s private language, are rarely the things we planned. They are the things that were true enough in the moment to be worth carrying forward. The pancakes were lovely, as they always are. But it was the popcorn song that made the weekend, and it made it because neither of us saw it coming.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c8fc05a00bff" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Capability We Traded Away]]></title>
            <link>https://thomasjreige.medium.com/the-capability-we-traded-away-0073d47b4df4?source=rss-0c9ba822d379------2</link>
            <guid isPermaLink="false">https://medium.com/p/0073d47b4df4</guid>
            <category><![CDATA[psychology]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[technology]]></category>
            <dc:creator><![CDATA[Dr. Thomas Jreige]]></dc:creator>
            <pubDate>Thu, 20 Aug 2026 06:34:20 GMT</pubDate>
            <atom:updated>2026-08-20T06:34:20.823Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jr3aHCt5V5LqVWUhKhY8iQ.png" /><figcaption>The Capability We Traded Away</figcaption></figure><p>I have been watching something for months now, and tonight it finally sat still long enough for me to name it. We are buying technology at a pace no generation before us has managed, and we have mistaken that pace for progress. What it has actually done, in home after home and organisation after organisation, is disrupt the very capabilities it was sold to enhance. Problem solving has thinned out. The willingness to sit inside a hard question until it yields has gone quiet. People, and increasingly children, expect to be handed the answer rather than to earn it.</p><p>I want to walk through why this is happening, because the mechanism is more precise than “screens are bad” or “kids these days.” It is a specific architecture, built for a specific purpose, and it does not stop working once childhood ends. It follows people into boardrooms.</p><h3>What the tantrum is actually telling you</h3><p>Start with the scene most parents recognise immediately. A device is taken away, and what follows is not a normal tantrum. It is something sharper, longer, and stranger, an intensity that does not match the size of the request. Parents who have lived through both describe the difference clearly: this is not the reaction to being told dessert is finished. This is something else.</p><p>Clinicians have started giving that difference a name. Research on preschool-aged children found that a meaningful share, roughly four in ten in one clinical sample, showed what researchers term withdrawal dysphoria, acute irritability specifically triggered by the interruption of device use, and reported by parents as distinct in intensity, duration, and emotional quality from their child’s usual outbursts. The researchers link this reaction to sustained activation of the brain’s threat and emotion circuitry, the same territory implicated in other compulsive behaviours. Whether the word “addiction” is the clinically correct term is still debated among specialists. What is not in dispute is the pattern itself, and any parent watching it happen at their kitchen table does not need a paper to confirm what they are seeing.</p><p>The other side of this is less dramatic and more worrying. It is the child who does not react at all. Flat. Unresponsive. Present in the room but not really in it. Not defiant, just absent, the way a person looks when the part of them that processes a request and weighs a consequence has gone quiet. Parents describe both states, sometimes in the same child within the same week, the explosion and the shutdown, and they are not two separate problems. They are the same mechanism showing two different faces.</p><p>That mechanism has a mundane origin. When screens become the default tool for calming a child down, the child does not learn to regulate difficult feelings on their own. They learn one strategy, and only one: soothe with the device. Their emotional toolbox ends up holding a single instrument. Take that instrument away and there is nothing else in the box, so what comes out is either rage or collapse, because both are what happens when a system has no other response available to it.</p><h3>Underneath the tantrum, a skill that never formed</h3><p>Behind the behavioural picture sits a developmental one, and it predates the tantrum by years. Research following children from infancy has found that heavier early screen use is associated with measurable delays in problem solving, communication, and fine motor skills, some of it visible as early as age one. Preschoolers with high screen exposure score lower on cognitive flexibility, the capacity to approach a problem more than one way, and on the specific, unglamorous skill of waiting. Waiting for a turn. Waiting for a toy. Sitting inside a small frustration instead of having it resolved instantly on their behalf.</p><p>That capacity to wait is not a minor behavioural footnote. It is the seed of everything that later becomes problem solving, patience under uncertainty, the willingness to sit with a question that does not yet have an answer. If a child never has to wait, because a screen removes the wait every time boredom or frustration appears, the seed does not get planted. What we are calling a discipline problem in a ten-year-old, or a resilience problem in a twenty-five-year-old employee, or a judgement problem in a forty-five-year-old executive, may in a great many cases be the same absence, showing up at different points along one continuous line.</p><h3>The architecture doing this is not an accident</h3><p>None of this is incidental to how these products are built. Friction removal, instant reward, and choice engineered down to a single frictionless tap are not side effects of good design. They are the design. It is the same architecture I describe elsewhere as the psychographic kill chain, the precision use of behavioural and psychological pressure points to move a specific person toward a specific outcome. The only real difference between that framework applied to an adversary targeting an executive and the same framework built into a children’s app is the age of the target and the stated intent of the builder. The mechanism does not know the difference. It was never designed to.</p><p>This is where the story stops being one about children and starts being one about all of us, because the reward architecture that captures a four-year-old’s attention is the same architecture, refined and pointed differently, that now sits inside the tools adults use to make decisions. The habit formed early does not disappear at eighteen. It graduates.</p><h3>Where this actually costs you</h3><p>Here is where I ask you to sit with the uncomfortable extension of this argument, because it applies directly to the room you are in, not just the one your children are in.</p><p>There is now a growing body of research on what happens to adults who lean heavily on AI tools for thinking rather than for information. One substantial study, surveying and interviewing hundreds of participants across age groups, found a clear negative relationship between frequent AI tool use and independent critical thinking ability, driven by what researchers call cognitive offloading, the quiet transfer of reasoning work from the person to the system. Younger adults show the strongest dependence. Higher education appears to offer some protection, but not immunity.</p><p>A separate strand of that same research points to something more dangerous for anyone who carries decision authority. Overreliance on confident, fluent AI output suppresses reflective evaluation. It miscalibrates a person’s confidence in their own judgement, particularly under time pressure or complexity, the exact conditions under which a leader’s judgement matters most. The result researchers describe is deskilled judgement, capability that has quietly eroded because it was never exercised, sitting undetected until the moment it is needed and is not there.</p><p>That is the version of the tech tantrum that never gets diagnosed in a boardroom. Nobody screams. Nobody collapses on the floor. What happens instead is a leader stands in front of a genuinely uncertain decision, the kind no system can pre-digest for them, and reaches for the reflex that has served them for years, ask the tool, trust the fluent answer, move on. Except this time the tool has nothing true to offer, because the situation is new and the stakes are real and there is no dataset that has seen this exact configuration of pressure before. The friction that should have been building judgement all those years was routed around instead. And in the moment that friction was finally required, it was not there to draw on.</p><h3>The single point of failure boards are not measuring</h3><p>Boards are reasonably good, these days, at asking about model risk. Bias in the training data. Regulatory exposure. Who is accountable when an automated system makes the wrong call. What almost no governance framework asks is a quieter and, I would argue, more consequential question: what has happened to the humans supervising these systems. What has been lost in the people who are meant to catch the model’s mistakes, override its bad recommendations, and exercise the judgement the system cannot.</p><p>Applied inward, the same discipline I use to map dependencies and cascading failure points in an organisation’s external risk exposure applies just as cleanly to its own workforce. The single point of failure a board should be worried about is not always a vendor or a piece of infrastructure. Sometimes it is a management bench that was trained entirely inside frictionless, assisted environments, arriving at senior decision points having never had to sit, unaided, inside genuine uncertainty. That is not a hypothetical for the future. Given how long these tools have already been embedded in how people learn and work, it is a bench that is being assembled right now, and it will be making decisions in rooms you sit in within the decade.</p><p>A human capability floor deserves the same governance status as a model risk register. Not a ban on the tools. A deliberate insistence that independent reasoning, the capacity to sit inside a problem before a system resolves it, is preserved as a matter of institutional discipline, not left to erode quietly because nobody made preserving it anyone’s job.</p><h3>The light going out</h3><p>There is a reason strategy only really reveals itself in the absence of certainty. Anyone can follow a lit path. What tells you whether a person or an organisation has real judgement is what happens when the light goes out, when there is no system fluent enough to hand over an answer, and the only thing left in the room is whoever is standing in it.</p><p>The child who has never had to wait for the toy, and the executive who has never had to sit inside an unresolved decision without reaching for a fluent tool, are further along the same line than either would like to admit. The convenience never announces its cost on a quarterly report. It shows up later, in a moment that actually matters, when the capability that should have been there simply is not, because it was never built.</p><p>We did not disrupt our lives by accident. We disrupted them by outsourcing, piece by piece, the very friction that used to make us capable. The question worth asking tonight is not whether the technology was worth adopting. It almost always was. The question is what you have quietly stopped practising because it got easier not to, and whether you would still recognise your own judgement if the tool in your hand went silent tomorrow.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=0073d47b4df4" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[When Security Rests on a Relationship, You Do Not Have Security]]></title>
            <link>https://thomasjreige.medium.com/when-security-rests-on-a-relationship-you-do-not-have-security-4e89fbac8ee2?source=rss-0c9ba822d379------2</link>
            <guid isPermaLink="false">https://medium.com/p/4e89fbac8ee2</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[psychology]]></category>
            <category><![CDATA[geopolitics]]></category>
            <dc:creator><![CDATA[Dr. Thomas Jreige]]></dc:creator>
            <pubDate>Wed, 19 Aug 2026 05:07:31 GMT</pubDate>
            <atom:updated>2026-08-19T05:07:31.996Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Z61XdaSdH-fPJiicq3m5zA.png" /><figcaption>When Security Rests on a Relationship, You Do Not Have Security</figcaption></figure><p>On 16 August 2026, the President of the United States announced on Truth Social that he had instructed the Secretary of War to substantially reduce the joint military exercises the United States runs with South Korea. The stated reason was not a change in the threat. It was a relationship. In his words, the drills send a signal that is hostile toward a country that, for as long as he has been President, has been unthreatening and respectful. The reference was to North Korea, and to Kim Jong Un.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/621/1*X1K_wgppZl55RoWVCiaRWg.png" /></figure><p>The coverage that followed did what coverage does. It argued about whether the decision was a concession, whether it weakened the alliance, and whose fault it would be if Pyongyang read it as weakness. Those are reasonable questions. They are also the wrong place to spend your attention if your job is to understand risk, because they treat the decision as the event. The decision is not the event. The decision is the evidence.</p><p>What it reveals is more useful than what it does. A defence commitment that had stood for decades, backed by treaty and force posture and a great deal of institutional machinery, was reduced in an afternoon on the strength of one leader’s read of another leader’s manner. Everything downstream of that fact is worth thinking about carefully, and very little of it is about Korea.</p><h3>The dependency you could not see until it moved</h3><p>Alliances behave like controls. You can describe them on paper in terms of what they are supposed to do, and you can believe that description right up until something applies pressure and shows you what they actually rested on. On paper, the posture on the Korean peninsula rested on obligations and capabilities. In practice, at the moment that mattered, it rested on one man’s sense of whether another man had been respectful.</p><p>This is the pattern we spend most of our time helping boards see in their own organisations, and it rarely announces itself. A control that depends on a single relationship is not a control. It is a hope with a signature attached. It works reliably while the relationship holds, which is exactly the condition that makes people stop examining it. Dependencies of this kind are invisible during good weather and total during bad, and the transition between the two states can happen faster than any governance process is built to absorb.</p><p>The Korea decision is a clean illustration because the dependency became visible at the same instant it moved. For years the commitment looked structural. In August it turned out to be personal, and it turned out to be personal all along. Nothing changed except that the pressure arrived and revealed where the weight had been sitting.</p><h3>What the adversary was actually building</h3><p>There is a version of this story where North Korea spent two decades trying to out-build the United States and failed, and a version where it did something more patient and more effective. The second version is the one worth studying.</p><p>An actor that cannot match you on force can still shape the variable your commitments are calculated against. That variable is judgement, and judgement is reachable. It responds to how understood a person feels, how respected, how much they believe a relationship is real and reciprocated. Being unthreatening and respectful is a posture, and a posture is a choice made in service of an objective. The objective here was never to win a confrontation. It was to make confrontation feel unnecessary to the one person with the authority to reduce the pressure.</p><p>We would put it plainly and then leave it alone. The phrase in the announcement, that Kim has been unthreatening and respectful, is now doing work as a reason for a policy. That is what a successful influence posture looks like from the outside. It does not appear as coercion. It appears as a relationship the target is proud of, cited approvingly, in the target’s own words, as the basis for a decision that serves the other side. Whether anyone intended to be moved is beside the point. The evidence that the approach worked is that it is being repeated back as justification.</p><h3>What the decision process gives away</h3><p>Set the geopolitics aside and read the announcement as a governance artefact, the way a regulator or a court eventually would. A significant allied-defence decision was communicated in public, framed by a personal relationship rather than by any assessed change in conditions, delivered alongside an unrelated remark about Iran, and partly justified by the observation that it was too late to cancel the exercises anyway.</p><p>For a board audience this is instructive, because it shows what an undefended decision looks like. There is no visible record tying the choice to a change in the threat environment. There is no rationale that anyone other than the decision-maker could reconstruct or stand behind. There is no chain of reasoning that survives the departure of the individual who made it. A directors’ duty is not measured by whether a decision felt right in the moment. It is measured by whether the decision was defensible when someone examines it later, without the benefit of the relationship or the mood that produced it. A decision that lives entirely inside one person’s rapport with a counterparty fails that test before anyone has even challenged the substance.</p><p>The uncomfortable part is how ordinary this failure mode is below the level of heads of state. Organisations approve acquisitions, renew critical suppliers, and set risk appetite on the strength of a trusted relationship, and they record the outcome without ever recording the reasoning. The record looks complete. It is complete right up until the relationship changes and someone asks why the decision was made, and finds that the answer left the building.</p><h3>The signal travels further than intended</h3><p>A reduction in exercises communicates, because exercises were always partly a form of communication. The intended audience was Pyongyang. The actual audience is every capital that has been quietly estimating how durable American commitments are. Seoul, Tokyo and Beijing do not need a policy paper to update their assumptions. They need one gesture, and they now have it.</p><p>This is the behavioural point generalised beyond this case. Your counterparties are continuously running a model of you, and they revise that model on evidence you did not think you were offering. Leaders consistently underestimate this, because they experience their own decisions from the inside, where context and intention are obvious, and they forget that everyone else experiences those decisions from the outside, where only the signal is visible. A single line, written quickly, can move other people’s picture of you more than a year of stated policy, and you will usually be the last to know it has happened.</p><h3>The exposure you have probably not priced</h3><p>The reason this decision is worth an organisation’s attention has nothing to do with Korea. It is that the same structure sits inside most institutions, unexamined, because it has never been tested.</p><p>Look at where your own security, continuity or leverage actually rests. Somewhere in the structure there is a relationship, a single person’s judgement, or a counterparty’s continued goodwill, carrying more weight than the org chart admits. While conditions are calm, that dependency reads as stability. It is not stability. It is unpriced exposure that will hold until the day it does not, and it tends to move on someone else’s timing rather than yours.</p><p>Two questions are worth sitting with. What in your structure would move if one relationship changed. And if it moved, who besides you could explain and defend the decisions that relationship had quietly been making.</p><p>The value of the Korea decision is that it is public, recent and impossible to dispute. Most organisations will never get a warning this clear about a dependency they cannot see. This one is on the record.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4e89fbac8ee2" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>