On June 12, 2026, the US government issued a directive and Canadian businesses lost access to two of the world's most advanced AI models overnight. No warning. No transition period. No appeal process. Just gone.
That should concern every business owner in this country who is building their operations on AI.
The models in question were Fable 5 and Mythos 5, developed by Anthropic, one of the leading AI companies in the world. The US government cited national security concerns and ordered Anthropic to immediately suspend access for all foreign nationals. Anthropic complied within hours. Canadian users, mid-workflow, were cut off because a foreign government made a decision that we had no voice in.
I want to be direct about something: this article is not about whether the US government was right or wrong. That is their debate to have. This is about what it means for Canadian businesses that we were collateral damage in it.
What Actually Happened
Anthropic published a detailed public statement explaining the situation. The government claimed to have evidence of a jailbreak, a method of bypassing Fable 5's safety guardrails, that could enable misuse related to cybersecurity.
Anthropic's response was measured but firm. They complied with the legal directive, but they also disagreed with the reasoning publicly. Their position: the jailbreak demonstrated was narrow and non-universal. It could surface only minor, already-known software vulnerabilities. The kind that security professionals use every single day to defend systems. They also noted that comparable results are achievable from other publicly available AI models already on the market.
Anthropic has been unusually transparent throughout this. Before Fable 5 launched, they worked with the US government, the UK AI Safety Institute, and multiple third-party organizations to red-team the model for thousands of hours. They stated publicly that perfect jailbreak resistance is not achievable for any AI model right now and that their strategy was defense in depth, not perfect immunity. When the directive came, they published their reasoning, named their disagreement clearly, and committed to sharing more details within 24 hours.
That is not the behavior of a company hiding something. But that is also not the point.
The point is that Canada had zero input into this decision, and Canadian businesses paid the price for it immediately.
Canada Was a Bystander to a Decision That Affected Us Directly
Think about what this looks like in practice. A law firm in Vancouver using Fable 5 for legal research. A manufacturer in Abbotsford using it to accelerate product development. A nonprofit in Surrey using it to stretch limited staff capacity. A healthcare organization using it to reduce administrative backlog. All of them, mid-workflow, shut out. Not because of anything they did. Not because of any Canadian regulatory decision. Because of an export control dispute between a US government agency and a US company.
We did not choose to be subject to this jurisdiction. We inherited it by default because we built our AI strategy on American infrastructure without building Canadian alternatives alongside it.
This is not a new problem. Canada has spent decades building economic and technological dependence on our neighbour to the south. In many areas, that has made sense. Proximity, shared language, integrated supply chains. But AI is different. AI is becoming the infrastructure that runs businesses. And infrastructure dependency is a sovereignty issue.
The businesses hit hardest yesterday were the ones with no contingency. No documented AI dependencies. No vendor continuity plan. No governance framework. They had adopted AI tools informally, operationalized them quickly, and never asked the question: what happens if this goes away? That question has an answer now. And it is not a comfortable one.
The Governance Gap Nobody Wants to Talk About
In my work with BC businesses on AI strategy, governance is always the conversation that gets deferred. Business owners want to talk about productivity gains. They want to know which tools to use. They do not want to talk about what happens when those tools disappear overnight or when a foreign government decides their access is a national security risk.
That conversation cannot be deferred anymore.
Every organization using AI tools needs to be asking a basic question: what is my continuity plan if my primary AI provider becomes inaccessible? That is not paranoia. It is the same risk management thinking we apply to any critical vendor dependency. After June 12, 2026, it is not a hypothetical anymore. It is a documented risk event.
The businesses that had AI governance frameworks in place were better positioned yesterday. They had documented their tool dependencies. They had data handling policies that did not lock them into a single vendor. They had started thinking about AI continuity the same way they think about backup and disaster recovery. The businesses operating without any governance structure woke up to a gap they did not know they had.
For organizations in regulated industries, law, healthcare, finance, government, this event should be a forcing function. The regulatory environment around AI is no longer abstract. It is operational. It has real business consequences. And it is moving faster than most Canadian organizations have been willing to move on governance.
Canada Needs to Be in This Game, Not Just Watching It
Canada has world-class AI research capability. We have the Vector Institute, Mila, and Amii. We have talent the rest of the world recruits aggressively. We are not without resources.
What we lack is the commercial infrastructure and the policy urgency to turn that capability into sovereign AI capacity. While Canada studies and panels and reports, the United States is deploying and, as of June 12th, controlling who gets access to that deployment. The EU is regulating. Other nations are investing at scale.
Canada keeps playing a long game in a competition that is already well underway.
I am not arguing for AI isolationism. That is neither practical nor desirable. What I am arguing for is intentionality. Canadian businesses should be thinking about AI vendor diversification the same way a smart operator thinks about supplier concentration risk. Canadian government needs to treat AI infrastructure as critical national infrastructure, because that is exactly what it is becoming.
We need Canadian-hosted options. We need governance frameworks that do not simply defer to US regulatory interpretations. We need procurement policies that treat vendor jurisdiction as a risk factor. And we need federal and provincial governments to move with actual urgency, not the pace of a standing committee.
What BC Businesses Should Do Right Now
This is not just a policy problem. There are concrete steps every organization should be taking today.
Audit your AI dependencies. List every AI tool your organization uses. Who provides it? Where is it hosted? What jurisdiction controls it? If you do not know the answers, that is your first governance gap.
Build a vendor continuity plan. For any AI tool that is operationally important, know your fallback if access is suspended. This does not mean running everything twice. It means having a plan before you need one.
Review your data governance. If your organization operates in a regulated industry, understand exactly what data is flowing through which AI tools and under which terms. Cross-border data flows to US-based AI providers carry compliance implications that need to be documented, not assumed.
Have the AI governance conversation with your technology provider. Not "how do we use AI tools." That conversation is already happening. The conversation most organizations have not had is "how do we govern our AI exposure as a business?" That distinction matters, and it matters now.
The Larger Lesson
Anthropic acted in good faith throughout this. They built strong safeguards. They were transparent about their limitations. They complied with a legal directive they disagreed with and said so publicly. None of that changes the outcome for Canadian users.
The continuity of our AI access depended on the outcome of a dispute between an American company and the American government. A dispute we had no voice in, no representation in, and no ability to influence.
That is the structural problem. Not Anthropic. Not even this particular government decision. The structural problem is that Canada has allowed itself to become entirely dependent on foreign AI infrastructure without building the domestic capability, governance frameworks, or policy environment to protect that dependency from exactly this kind of disruption.
The AI era is here. The question Canada has to answer, and has to answer faster than we have been moving, is whether we are going to be architects of our own AI future or tenants in someone else's.
Yesterday was a reminder of what it looks like to be a tenant.
Darren Coleman is the Founder and CEO of Coleman Technologies Inc., a Langley-based managed IT, cybersecurity, and AI strategy firm serving organizations across Greater Vancouver and the Fraser Valley since 1999. Coleman Technologies helps business leaders adopt AI safely, strategically, and with clear governance in place.
