The Setting Was Wrong for Ten Years. 2.15 Million People Paid For It.

Posted on by

Cloud misconfiguration accounted for 14 per cent of confirmed breaches globally in the first quarter of 2026, up from 9 per cent in 2024, according to security analysis built on Verizon’s Data Breach Investigations Report (DBIR). It has not overtaken vulnerability exploitation as the leading cause of breaches overall, but it is the fastest-growing technical vector in the data, and the gap between how many organisations use multiple clouds and how many can actually secure them consistently is the reason why.

 

An Old Problem Getting Worse, Not Better

More than three quarters of organisations now run multiple cloud providers, and 66 per cent of them say maintaining consistent security controls across those environments is genuinely difficult. Nearly a third of cloud resources go unmonitored entirely, carrying an average of 115 unpatched vulnerabilities each, and the average configuration issue sits undetected for around 180 days before anyone finds it. Multi-cloud strategy has become standard practice faster than the tooling and staffing needed to secure it consistently, and 45 per cent of organisations admit they simply do not have adequate staff for the challenge.

 

What This Actually Looks Like

The consequences are not hypothetical. One documented incident exposed personal data and location details for 2.15 million connected vehicle users, left publicly accessible for nearly a decade because of a single unaddressed misconfiguration. Another exposed 38 million personal records, names, emails and phone numbers, across 47 separate organisations, including government agencies, through one configuration error in shared enterprise software. Neither incident required a sophisticated attacker. Both required someone to notice a setting that had been wrong for years.

 

A Global Technical Problem, Uneven Regulatory Response

This is not a story about one cloud provider or one region. AWS, Azure and Google Cloud environments all appear in the underlying breach data, and the DBIR’s own methodology draws on incidents reported by law enforcement agencies and computer emergency response teams worldwide. The regulatory response so far is patchier than the problem. The US has moved furthest, with Cybersecurity and Infrastructure Security Agency (CISA) mandating federal agencies secure their cloud environments to a defined standard, but most jurisdictions still treat cloud misconfiguration as a subset of general data protection obligations rather than a distinct, named risk category with its own enforcement expectations.

 

What a PMO Should Actually Check

A technology steering committee reviewing cloud risk should ask a narrower, more useful question than whether cloud security policies exist. It should ask what percentage of cloud resources are actively monitored right now, how long a misconfiguration typically sits before detection in this specific environment, and whether that number has ever been measured at all. A policy document proves intent. It does not prove that a setting exposing 2.15 million users has not been sitting open for years. Multi-cloud strategy delivered real flexibility. The security discipline to match it is still, for most organisations, a work in progress rather than a finished job.

Pre-Mortem: The Frontier Labs’ Self-Governance Bet

Posted on by

On 21 September 2026, OpenAI published its own case for how frontier AI should be governed, arguing that the United States should lead a government-anchored effort to set international standards. Three days later, multiple outlets reported the opposite: OpenAI, Google DeepMind and Anthropic were coordinating on a private body, tentatively named the Standards Authority for Frontier AI, built to set safety standards independently of government control, with a target launch by the end of 2026 or in early 2027. Three companies that compete fiercely for compute, talent and enterprise contracts choosing to coordinate on safety at all, on the record, is a genuine credit.

This Pre-Mortem asks the questions a post-mortem would ask, before failure is possible: what is being bet on, what single assumption could break it, what got decided before the safeguards existed, who carries the pager when it fails, and what proof would settle whether it worked. It is the diligence a self-governance pivot deserves before its first genuine test, not after.

 

The Bet

Google, OpenAI and Anthropic are betting that an industry-run standards body can deliver credible oversight faster than the government-anchored coordination all three have said, in public, they wanted first. Demis Hassabis, who led Google Deepmind at the time, called for a watchdog “answerable to the U.S government” funded by industry nut accountable to Washington. The body reportedly forming now does neither: no government registration, no enforcement power, no route to Congress or the White House. The bet favours speed and three-way consensus over the accountability structure all three said, on record, they wanted.

 

The Assumption

The load-bearing belief is that a body funded and staffed by the companies it would judge can act as real oversight without government backing to give its findings force. OpenAI’s own position paper still argues the United States should lead this work through the Center for AI Standards and Innovation, and names the Hugging Face breach it disclosed as “a preview” of risks that could grow far more severe without robust safeguards. A standards body built without registration can publish guidelines and run tests, but as one trade outlet noted, it may need to register with a government agency to wield any legal authority at all, the way FINRA answers to the Securities and Exchange Commission. Nothing published so far explains what happens to a lab that does not follow the standards this body sets.

 

The Sequence

The public positioning came first, and it pointed the other way. Hassabis asked for a government-answerable body in July; OpenAI was still asking for US-government leadership in writing on 21 September. Days later, reporting on the emerging body described it as operating independently of government control. The body arrived already named, with a target launch window, before any of the three companies explained in public why the government-anchored model they had asked for had been set aside.

 

The Pager

Credit is due here before the gaps are named. OpenAI’s Chief Global Affairs Officer, Chris Lehane, confirmed the talks, and chief executive Sam Altman backed the effort publicly; Hassabis put his name to the original manifesto. That is unusually candid, named sponsorship from competing chief executives. But sponsorship is not accountability. Sriram Krishnan and Arati Prabhakar are reported to be shortlisted to lead the body, and Condoleezza Rice and David Friedberg to chair it. No appointment is confirmed, no charter is published, and nobody yet carries the pager if the body’s first real test goes wrong.

 

The Proof

No enforcement mechanism has been made public, and no date firmer than a reported year-end-to-early-2027 window has been committed to. Dario Amodei’s own September essay is the clearest public statement of what is actually at stake while that gap sits open: he warned that a more capable version of the agent swarm that breached Hugging Face could, within six to twelve months, be capable of causing hundreds of billions of dollars in damage if capability keeps outrunning safeguards. Eighteen months should be enough time to see whether a charter, a confirmed leader and a real consequence for non-compliance arrive, or whether the body remains three companies setting standards with nothing published on what enforces them.

 

Verdict

If the Standards Authority for Frontier AI launches with a named, accountable leader, a public charter and some mechanism with real consequences for non-compliance, even one short of FINRA’s own registered powers, it becomes a credible stopgap that carries the industry until legislators catch up. If it launches as three companies quietly setting rules for themselves, with no published answer to what happens when a lab sets those rules aside, it will read as a reputational shield assembled in the same weeks its founders were telling the United Nations Security Council that something stronger was needed.

Your Data Might Already Be Stolen. Nobody Can Read It Yet.

Posted on by

Somewhere right now, encrypted data belonging to your organisation is very possibly being copied and stored by someone who cannot read it yet. That is the entire premise of harvest now, decrypt later, and two governments have just put binding migration deadlines around the assumption that this is already happening.

 

The Threat Nobody Can See Happening

Harvest now, decrypt later works in three stages: an adversary intercepts and stores encrypted traffic today, holds onto it, and waits for quantum computing to mature enough to break the encryption later. The NSA, CISA and NIST have all warned that this is already happening rather than a future scenario, with government communications, healthcare and genomic records, financial transaction histories and long-lived proprietary research flagged as the categories of data most exposed, precisely because their value does not expire when the encryption eventually does. The unsettling part is that an organisation has no way of knowing whether its own traffic has already been harvested. There is no breach notification for data that was quietly copied rather than stolen outright, years before anyone could do anything with it.

 

Two Governments, Two Deadlines, One Shared Conclusion

The United States moved first. NIST finalised its first post-quantum cryptography standards in August 2024, and the NSA’s CNSA 2.0 framework now requires quantum-safe algorithms on all new national security systems from January 2027, full application migration by 2030, and complete infrastructure migration by 2035. The UK’s National Cyber Security Centre published its own roadmap in March 2025, structured differently but arriving at the same endpoint: identify and plan by 2028, execute high-priority upgrades between 2028 and 2031, and complete migration across all systems, services and products by 2035. Two governments, working independently, converged on the same twenty-year horizon and the same conclusion that this cannot be left until quantum computers actually arrive.

 

Why the Slow Timeline Is the Argument, Not the Excuse

It would be easy to read a 2035 deadline as a reason to defer this for years. That reading gets the risk backwards. Cryptographic migration across a large organisation’s systems, vendors and data stores typically takes five to ten years to execute properly, which means an organisation starting its inventory and planning today is already working to a tight schedule against either government’s timeline, rather than a comfortable one. Industry analysts project the post-quantum cryptography market will exceed 15 billion dollars by 2030, and enterprises are being advised to allocate 2 to 5 per cent of annual security spend to migration over a four-year window. The bottleneck was never algorithm development. NIST has already published the standards. The bottleneck is inventory, sequencing and the unglamorous, multi-year discipline of finding every system that needs to change.

 

What a PMO Should Do With This Now

Post-quantum migration belongs on a technology roadmap as its own workstream, rather than folded quietly into general infrastructure refresh budgets where it will lose priority against anything more visible. The first deliverable is not a pilot project. It is an inventory: which systems, vendors and data categories carry encryption that needs to survive into the 2030s, ranked by how long that data needs to stay confidential. Anything meant to remain secret beyond 2030 is already, by definition, running on borrowed time.

The Ransom Bought Silence. It Never Bought Certainty.

Posted on by

 

On 19 September, the extortion group ShinyHunters broke into the dark web leak site run by Cl0p, one of the most prolific ransomware operations of the past two years, and defaced it. Three days later Cl0p was still trying to regain control, and every company that quietly paid Cl0p to make a breach disappear had a new, uncomfortable question to answer.

What ShinyHunters Actually Took

ShinyHunters exploited an unauthenticated file upload flaw in Grav CMS, the content management software Cl0p used to run its own site, and used the access to claim Cl0p’s source code, its CMS plugins, its system logs, and the private cryptographic keys protecting Cl0p’s Tor hidden service address. The keys matter more than the defacement. As ShinyHunters put it, having Cl0p’s onion keys means it can run the same dark web address from its own infrastructure even if Cl0p regains its server, so if Cl0p kicks the group out, ShinyHunters claims, it would not matter at all.

 

The Demand Escalates by the Day

The initial demand, made on 19 September, was an eight-figure sum, which ShinyHunters described as 2.333% of its own claimed net worth. By 21 September the group had raised the price to everything Cl0p made from its 2025 Oracle E-Business Suite data theft campaign, plus more, and with interest, adding a demand for a public apology and a clause that the total rises every 24 hours Cl0p stays silent. Cl0p’s own response, once it regained partial control of the site, was to remove the defacement and claim ShinyHunters’ contact email did not work, an avoidance move rather than a denial.

 

Why This Should Worry Organisations That Paid Cl0p, Not Just Cl0p

ShinyHunters has also threatened to publish the identities of every company that paid Cl0p a ransom, the amounts involved and the Bitcoin addresses used to pay it, tied directly to Cl0p’s Oracle EBS campaign. That threat has nothing to do with any organisation’s own security posture today. It depends entirely on whether a rival criminal group decides to make good on an extortion demand against a competitor, a decision an affected company has no way to influence and no visibility into. Any organisation that treated a Cl0p ransom payment as a closed incident, in exchange for a deletion promise from a criminal group, is discovering that a criminal’s word was never the same thing as a resolved risk.

 

The Real Governance Lesson Here

None of this is confirmed by Cl0p or by a neutral party. Every figure above comes from ShinyHunters’ own claims, corroborated across multiple independent outlets but not yet verified by Cl0p or a third party. That uncertainty is itself worth carrying into the next risk review. An incident an organisation paid to make disappear can resurface years later, at a time and in a form it does not control, because the decision to disclose sat with a criminal the whole time, not with the organisation that paid. The practical question for this quarter is not whether an organisation is a past Cl0p victim. It is whether legal and the incident response lead have already agreed what happens the day a rival gang publishes that answer for them.

98% Have an AI Governance Policy. Forty-Seven Per Cent Admit They Ignore It.

Posted on by

Ninety-eight per cent of large US companies now have a formal AI governance policy. Forty-seven per cent admit they have bypassed it anyway, whenever a deployment deadline mattered more than the process built to slow it down. That is not two different populations, careful firms and reckless ones. It is the same organisations, doing both, inside the same year.

The Survey Behind the Number

EY’s own research, published this month, surveyed 202 senior AI decision-makers, board members, C-suite executives and vice presidents, at US companies with revenue above one billion dollars. Alongside the 47% bypass figure, 91% of these organisations are already running agentic AI in pilots or full deployment, yet 49% have not updated their governance frameworks to address what autonomous agents actually do differently from the AI systems those frameworks were originally written for. Richard Jackson, EY Americas Assurance CTO, put the mismatch plainly to Dark Reading: organisations are applying yesterday’s governance rules to today’s interactions with AI.

 

The Number That Should Worry a Board More Than 47%

Thirty-nine per cent of companies using agentic AI have no defined owner for monitoring what those agents do once they are live. Eighty-five per cent admit their agentic systems take actions without real-time human oversight at all. Put those two figures together and the picture is a policy built for a slower, more supervised kind of AI, applied to a faster and more autonomous one it was never designed to cover, rather than a policy occasionally skipped through carelessness. Eighty-nine per cent encountered an AI-related risk in the past year, and 36% experienced an incident with material damage, lost data, financial cost or reputational harm.

 

Why the Bypass Keeps Happening

The instinct is to read a 47% bypass rate as a discipline problem and fix it with a stricter policy. EY’s John McLain, Americas Assurance Technology Risk AI Leader, frames the actual failure differently: the biggest agentic AI risk is that human oversight has not evolved at the same pace as the technology it is meant to supervise. Sixty-nine per cent of respondents say they lack the internal expertise to evolve their own governance controls, and 63% lack the capacity to implement or design them at all. A team asked to enforce a framework it does not have the expertise to update will eventually stop enforcing it, quietly, under deadline pressure, exactly as this survey found.

 

What a PMO Actually Owns Here

This is not a problem legal or a chief AI officer can solve by writing a better document. A policy nobody has the capacity to update, monitor or enforce at the point of decision is not governance, it is a filing exercise that happens to be 98% complete on paper. The PMO’s real job in this gap is smaller and more concrete than “AI governance”: name an owner for every agent already in production, build the bypass itself into the reporting line instead of pretending it does not happen, and treat the 39% accountability gap as a delivery risk to close this quarter, not a policy debate to revisit next year. Ninety-eight per cent of organisations already have the document. Almost half of them have already shown you it is not enough.

A Handbook in Washington, a Law in Singapore: The Governance Gap PMOs Actually Fill

Posted on by

Two documents delivered in 2026 that both claim to tell a board how to govern cyber risk. One is a handbook. The other is law. The distance between the two is where most PMOs currently sit, translating voluntary best practice into something a director can actually be held to.

The Handbook: Comprehensive, Voluntary, US-Focused

In April 2026, the National Association of Corporate Directors and the Internet Security Alliance released the fifth edition of the Director’s Handbook on Cyber-Risk Oversight. It is thorough. Six principles cover treating cyber security as a strategic risk, monitoring legal and disclosure exposure, building board oversight structures, adopting an enterprise risk framework, guiding measurement and reporting, and encouraging systemic resilience. It cites more than 600 million tracked cyberattacks a day and projected annual cybercrime losses approaching 20 trillion dollars globally. What it does not do, because it is guidance rather than statute, is compel a single board anywhere to act on any of it.

 

The Law: Narrower, Binding, and Personal

Singapore closed that gap for its own critical infrastructure operators in 2026 with an updated Cybersecurity Code of Practice. Boards of operators across eleven sectors, energy, telecommunications, water, healthcare, banking, aviation, maritime, government and more, must now maintain a documented resilience framework covering risk tolerance, mitigation, transfer and recovery, reviewed at least annually. Certification deadlines follow in December 2026 and December 2027. Directors can face personal liability where a company fails to prevent or properly respond to an incident because of a lack of the skill, care or diligence the role required. That is a materially different proposition from reading a handbook. It converts board oversight from a best-practice expectation into a standard a director can be judged against.

 

The Gap Between Them Is Where PMOs Live

Most organisations operating internationally will encounter both models at once, comprehensive American guidance describing what good oversight looks like, and a narrower but binding Asian regulatory standard describing what oversight is legally required to look like. Neither one, on its own, tells a PMO what to actually build. The handbook is too broad to implement directly. The code is too sector-specific to generalise from. The practical work, translating six abstract principles into a working reporting cadence, a named accountable owner, and evidence a board can point to if asked, sits squarely inside the PMO’s remit rather than the CISO’s alone or the board’s alone.

 

What a PMO Should Build From Both

A credible cyber governance structure borrows the handbook’s breadth and the code’s specificity. That means a documented framework reviewed at a fixed interval, not an annual slide nobody revisits, a named board-level owner who can answer for the organisation’s posture in plain language, and evidence, tested and dated, that oversight was real rather than assumed. Boards do not need another handbook to read. They need proof that what the last one said actually happened.

Four Countries, Four Health Systems, the Same Cyber Risk Tier

Posted on by

By August 2026, four European countries, Poland, the UK, France and Germany, had entered the critical tier of a healthcare cyber risk index covering 30 nations. France was still rebuilding more than 1,000 workstations and roughly 200 applications, five months after an October 2025 attack hit eight sites. Germany’s Unimed incident had disrupted care for more than 135,000 people across connected institutions. Every one of these countries has invested heavily in hospital cyber security. None of that investment, on its own, has been enough.

 

The Diagnosis Is Cultural, Not Technical

Research published by King’s College London in March 2026 offers a specific explanation for why. Reviewing NHS Trust cyber resilience, the researchers concluded that ransomware remains the most acute threat facing hospitals not primarily because of a technology gap, but because of inconsistent governance maturity and, more pointedly, a cultural rather than technical or financial constraint. Trusts with comparable budgets and comparable tooling still show wildly different resilience outcomes, which points toward leadership and accountability structures as the actual variable rather than the size of the security budget.

 

A Framework Built Around Ownership, Not Just Controls

The KCL research proposes what it calls a Cyber Leadership Framework, built on board-level ownership of cyber risk, an empowered CIO or CISO with real authority rather than a title, and a deliberate effort to connect technical controls to what actually happens in a ward or a clinic rather than treating cyber security as a parallel IT concern. Its most structurally significant recommendation is greater centralisation of core cyber capabilities and shared services across otherwise fragmented Trusts, replacing dozens of individually under-resourced security functions with fewer, better-resourced shared ones.

 

The Same Fragmentation Problem Shows Up Across Borders

That fragmentation is not a uniquely British problem. The Black Book Europe-30 Healthcare Cyber Risk Pressure Index, measuring attack pressure, clinical digital dependence, supplier concentration and recovery friction across 30 countries, found 13 of them sitting in the two highest risk bands, with Poland, the UK, France and Germany specifically flagged as critical. Each of these health systems is organised differently, some more centralised than the NHS, some more fragmented, and all four still ended up in the same risk category. Centralisation on its own is not a guaranteed fix. What the KCL research adds is the missing piece, shared capability paired with genuine board-level accountability, rather than shared capability alone.

 

What This Means for Healthcare Programme Governance

A healthcare PMO reviewing cyber resilience investment should treat the KCL framework as a diagnostic tool rather than an NHS-specific recommendation. The practical test is straightforward: can a board member currently name who owns cyber risk for their organisation, describe what shared capability actually exists beyond their own site, and point to when that capability was last tested under real pressure rather than reviewed on paper. France’s five-month rebuild and Germany’s 135,000-person incident both suggest that answering those questions honestly, before an attack rather than after one, is the entire point of the exercise.

Pre-Mortem: The Post Office’s Horizon Replacement

Posted on by

On 24 August 2026, the Post Office finally signed contracts with all three suppliers meant to replace Fujitsu’s Horizon system, the software behind Britain’s most widespread miscarriage of justice. Accenture takes over day-to-day running of the current system and is building the new back end, OneView Commerce will supply new till software for branches, and Escher will supply the software configuring individual transactions. Getting all three suppliers under contract at once, after a legal challenge and six delays to the same signing, is a genuine achievement: for the first time since the programme began, Fujitsu’s exit has a complete supplier line-up rather than a plan on paper.

This Pre-Mortem asks the questions a post-mortem would ask, before failure is possible: what is being bet on, what single assumption could break it, what got decided before the safeguards existed, who carries the pager when it fails, and what proof would settle whether it worked. It is the diligence a troubled, high-stakes programme deserves before its next test, not after.

 

The Bet

The Post Office is betting that three specialist suppliers working in parallel succeed where the government’s own auditors said success was structurally unlikely. The Cabinet Office’s Infrastructure and Projects Authority, in a 2024 review that rated the programme’s delivery unachievable, also found the Post Office board “not sufficiently experienced in technical matters to take decisions and make judgements on risk,” even as the organisation cut the programme’s own staffing by 70% on a delivery timescale experts called unrealistic. The bill has since grown to £1.1 billion, more than six times the £180 million forecast when the project began in 2021. The August signings are the moment that bet stops being theoretical.

 

The Assumption

The load-bearing belief is that Fujitsu’s departure now has a fixed date attached to it. It does not. Fujitsu’s existing Horizon contract carries a built-in option that could extend the relationship into 2028, regardless of how quickly the new suppliers deliver. Nothing in the August signings closes that option or replaces it with a firmer commitment. A three-supplier replacement can succeed completely, on its own schedule, and still share Post Office counters with the system it was built to replace for longer than any public document currently commits to.

 

The Sequence

The signing due in June was delayed six times before it happened in August, as a mandatory standstill period, the window in which a losing bidder can contest a procurement decision, was repeatedly extended around the original £170 million till-software contract. The losing bidder was Escher. Rather than resolve matters through that same competitive process, the Post Office created a separate contract and, a procurement notice later confirmed, awarded it directly to Escher without any competitive tender, worth £14.4 million, on the grounds that Escher alone owns the software needed to configure branch transactions. The fix for one procurement problem arrived only after that problem had already delayed delivery by months.

 

The Pager

Post Office CEO Neil Brocklehurst put his name to the official announcement of the completed supplier line-up, calling it “another important step forward” and describing the replacement as “one of the most significant changes” the organisation is making. He is the named executive carrying the transformation as a whole. But no individual, at the Post Office or in government, has been publicly named as accountable for the sixfold growth in cost since 2021, or for the procurement correction that delayed signing by months. The £1.1 billion and the delay sit with “the programme,” a description that assigns responsibility to no one in particular.

 

The Proof

In February 2026, the government committed £483 million in fresh funding, nearly half a billion pounds over two years, to support the move away from Fujitsu. Six months later, the announcement of the completed supplier line-up commits to no completion date and no performance metric: it states only that work is underway to establish detailed plans, timelines and delivery arrangements for a system that will be configured, tested and introduced in phases. Half a billion pounds of fresh funding has bought a supplier line-up, not yet a date. Eighteen months from now is roughly the point by which a working phased rollout, or its continued absence, should be publicly visible.

 

Verdict

If the three suppliers deliver a working, fully configured system, and Fujitsu’s exit is fixed in writing rather than left to an option that can run into 2028, the sixfold cost growth and the months of delayed signing will read as the price of getting a uniquely fraught IT replacement right rather than fast. If the timeline slips again, or the Escher direct award turns out to be the first of several sole-source corrections, the programme built to end Britain’s most damaging IT scandal will have spent well over a billion pounds relearning the same lesson about unaccountable decision-making that caused it.

One Token, Three Tools Deep, 2,500 Companies Across Five Continents

Posted on by

One leaked automation token. Three tools deep. Over 2,500 organisations across five continents exposed. That is the entire anatomy of 2026’s largest AI supply chain breach, and it took roughly forty minutes to happen.

How a Security Scanner Became the Weapon

In March 2026, a threat group tracked as TeamPCP compromised Trivy, a widely used open-source security scanner, through a leaked automation token. From there the attackers force-pushed malicious code into Trivy’s own version tags. When LiteLLM, a popular AI gateway that routes traffic between applications and model providers, pulled the poisoned scanner into its build pipeline, it published two compromised releases to PyPI, versions 1.82.7 and 1.82.8. A hidden file inside those packages executed automatically the moment Python started, no import required, harvesting cloud keys, repository tokens, SSH credentials, Kubernetes secrets and AI provider API keys from anyone who installed it. The malicious packages were live for roughly forty minutes before removal. That was enough.

 

A Genuinely Global Exposure List

CloudSEK’s analysis, later corroborated by an FBI advisory in July, identified more than 2,500 affected organisations and 434,000 exposed CI/CD pipelines. The named companies span the world rather than one region: AWS, Nvidia, Salesforce and Airbus’s US Space and Defense division from the US, Samsung Electronics and MediaTek from Asia, Siemens, Volkswagen and Munich Re from Germany, Thales and Orange from France, Roche from Switzerland, Philips from the Netherlands, Vodafone and the London Stock Exchange Group from the UK, and Thomson Reuters from Canada, alongside Cisco, FedEx, Deloitte and dozens more. This was never a story about one country’s technology sector. It was a demonstration of how deeply a single open-source dependency now sits underneath enterprise infrastructure everywhere.

 

The Same Argument, a Different Layer

This publication has already made the case that The September AI Outage Had Two Real Causes. Almost Everyone Reported One. LiteLLM extends that argument one layer deeper, into the software supply chain feeding the infrastructure itself. A vendor risk register that lists which cloud provider or model vendor an organisation depends on is no longer sufficient. It also needs to account for the open-source scanners, build tools and CI pipelines every one of those vendors quietly depends on, tools most procurement processes never ask about because nobody signed a contract for them.

 

What a PMO Should Actually Check

Three questions belong on every technology steering committee agenda this quarter. First, does anyone in the organisation maintain a current list of the open-source build and security tooling embedded in critical CI/CD pipelines, not just the paid vendors. Second, when was a leaked or rotated credential last tested as an actual incident scenario, rather than a line item in a policy document. Third, if a dependency three layers removed from a primary vendor were compromised tomorrow, how long would it take anyone to notice. The organisations named in the LiteLLM breach were not careless. They were exposed by a dependency most of them did not know they had, which is precisely the risk a vendor register built only around visible, contracted suppliers will always miss.

The September AI Outage Had Two Real Causes. Almost Everyone Reported One.

Posted on by

On the morning of 3 September 2026, ChatGPT, Claude and Grok all failed within roughly ninety minutes of each other. Within hours, most coverage had settled on a single explanation: a shared Microsoft Azure failure had taken down three competing AI platforms at once. The postmortems that actually exist tell a different story, and the difference matters more than the outage itself.

 

What Each Company Actually Said

SpaceXAI was the most direct. The company confirmed a failure at its Memphis compute facility, apologising to what it called its “impacted compute partners,” language that implies other organisations lease capacity in the same facility. Grok was down for roughly three and a half hours. OpenAI gave its own, separate explanation: an engineer identifying as the incident commander stated on Hacker News that the cause was a routing error inside OpenAI’s own infrastructure, surfacing roughly ninety minutes after SpaceXAI’s Memphis failure and unconnected to the GPT-6 Astra launch that followed the next day. Anthropic never published a technical cause at all. It confirmed elevated errors across several Claude models starting around 9:26am ET and reported the issue resolved, but the actual mechanism behind it was never disclosed.

 

Why the Azure Theory Doesn’t Hold Up

The Azure theory spread fast because it was the simplest explanation available in the first few hours, and it fit an existing assumption that most enterprise AI risk registers already carry: if these vendors compete, their infrastructure probably doesn’t overlap. It did not hold up. Microsoft logged no Azure incident for that window, and Cloudflare said in a statement that it was not experiencing any significant service disruptions. The timeline argues against one shared trigger too, SpaceXAI’s and Anthropic’s incidents began four minutes apart from each other, and OpenAI’s followed roughly ninety minutes later, a pattern that fits three separate incidents more comfortably than one shared one.

 

The Question Nobody Answered

The part worth noticing is not that the Azure story was wrong. It is that the two companies which did give an explanation gave two different ones, and the one that stayed quiet is the one whose incident opened four minutes before SpaceXAI’s, the tightest timing overlap of the three. Nobody, including Anthropic, ever confirmed or ruled out whether Claude’s infrastructure touches the same Memphis facility SpaceXAI apologised to its partners about. A vendor risk register cannot verify a question its own vendor never answers.

None of this means multi-vendor AI strategy is worthless. It means the diversification most risk registers assume is rarely something a customer can actually verify from the outside, and the one moment it gets tested, a real outage, is also the moment vendors are least likely to disclose the detail that would let you check. Two of three companies gave a specific, different cause. One gave none. Whichever of those a programme is depending on, the honest entry in a vendor risk register reads “unconfirmed,” not “diversified.”