22 Jun 2026 5 min read AI

Security Experts Fight US Export Ban on Anthropic AI

The US Commerce Department's export ban on Anthropic's Fable 5 and Mythos 5 has stripped defenders of critical AI tools and drawn a unified rebuke from over 100 cybersecurity professionals. Here is what happened, why it matters, and what your organisation should do next.

On 13 June 2026, the US Commerce Department issued an export control directive that forced Anthropic to abruptly disable its two most advanced AI models — Fable 5 and Mythos 5 — for all users worldwide, including those inside the United States. Over 100 cybersecurity leaders have since signed an open letter calling the move "dangerous." This is not merely a trade dispute. It is a fight over who controls the future of digital defence.

What Happened

Anthropic announced on 12 June 2026 that the US government had issued an export control directive ordering the suspension of all access to its newest frontier models, Claude Fable 5 and Mythos 5, which had been released just days earlier. The directive, issued by the Commerce Department, bars distribution of the models to any foreign national — not just individuals located outside the US, but also non-citizen employees working inside the United States, including Anthropic's own staff.

Given the global, interconnected nature of Anthropic's workforce and user base, the company determined the only viable compliance path was to disable the models entirely for everyone. The breadth of this restriction is worth underscoring: a directive ostensibly aimed at foreign adversaries effectively stripped access from American defenders as well.

The Professional Response

By 15–16 June, a coalition of more than 100 cybersecurity professionals — including well-known industry veterans — published an open letter addressed directly to Commerce Secretary Howard Lutnick and National Cyber Director Sean Cairncross. The letter argued that the export control directive "has taken the best models away from defenders, created market uncertainty, and risked America's AI leadership without any real risk to justify it."

This represents an unusually large and unified professional rebuke of a government policy. Signatories included prominent figures from across the security research community, and the letter received widespread coverage from TechCrunch, Dark Reading, SC Media, Infosecurity Magazine, and Cybersecurity Dive.

Questions of Motive

The Register reported on 22 June that the situation is growing more politically complicated. Internal Anthropic communications suggest the Trump administration may be using export controls as leverage over the company rather than responding to a genuine, documented national security threat. Anthropic itself noted that rival AI providers' models demonstrate comparable capabilities in uncovering code vulnerabilities, raising pointed questions about why Fable 5 and Mythos 5 were singled out.

If competing models offer similar capabilities, the national security rationale for targeting these specific models appears difficult to sustain on technical grounds alone.

Why It Matters

Defenders Lose Their Best Tools

For the cybersecurity community, Fable 5 and Mythos 5 were not merely impressive demonstrations — they were actively being used by security researchers to find vulnerabilities, audit code, and accelerate defensive work. Pulling those tools away from defenders whilst adversaries in other nations continue developing and deploying equivalent or competing models does not make the US safer; it arguably hands an asymmetric advantage to attackers.

As the open letter put it:

"To pull the best capabilities away from defenders while adversaries continue to develop and deploy them is not security — it is self-sabotage."

Geopolitical Fallout

The diplomatic consequences are equally significant. Al Jazeera reported that the ban has strained US alliances, with European officials reacting sharply. French Interior Minister Retailleau declared that nations must "master [AI] or suffer it," whilst former UK security minister Tom Tugendhat warned that "sovereignty is more about code than cannons."

The incident is accelerating conversations in allied nations about reducing dependence on US AI infrastructure — an ironic outcome for a policy supposedly designed to protect American technological leadership. As discussed earlier, if the directive's true purpose is competitive leverage rather than security, this geopolitical blowback represents a particularly costly miscalculation.

What You Should Do

If you are a security professional, IT leader, or part of an organisation that relied on Fable 5 or Mythos 5, the following steps are worth considering immediately:

Audit your AI-assisted security workflows. Identify which processes depended on Fable 5 or Mythos 5 and assess the gap. Document the operational impact — this data matters both for internal continuity planning and for potential policy feedback submissions.
Evaluate alternative frontier models. Competing models from other providers have demonstrated comparable code-auditing capabilities, per Anthropic's own statements. Test them rigorously, but ensure proper security vetting of any new AI tool before deploying it in sensitive workflows.
Monitor the policy situation actively. The Commerce Department directive is not permanent legislation. Follow GovInfoSecurity, SC Media, and Cybersecurity Dive for updates. The open letter campaign is ongoing and may yet produce results.
Consider signing or supporting the open letter. If you are a credentialled security professional, the letter remains open for additional signatories. Collective professional voices have historically moved export control policy — the Wassenaar Arrangement (an international framework governing the export of dual-use technologies) debates of the 2010s offer a relevant precedent.
Brief executive leadership and legal/compliance teams. If your organisation operates internationally or employs non-US citizens, this incident is a preview of how broadly export controls can be applied to AI tools. Build export-control risk into your AI procurement and continuity frameworks now.
Engage your government representatives. US-based professionals can contact their congressional representatives directly. International professionals can engage through their national cybersecurity agencies to apply diplomatic pressure through proper channels.

The Bigger Picture

This episode is not an isolated policy misstep — it is arguably a preview of the AI sovereignty battles that will define the next decade of cybersecurity. Governments worldwide are increasingly treating frontier AI models the way they once treated nuclear technology or cryptographic algorithms: as strategic national assets to be controlled, not shared.

However, the problem is that unlike nuclear warheads, AI models are software, and the knowledge embedded in them is rapidly replicated by competitors. History offers a cautionary tale: the US government's overly broad export controls on cryptography in the 1990s delayed the adoption of strong encryption domestically and internationally, ultimately weakening global security rather than strengthening it.

The security community's pushback on the Fable/Mythos ban is an attempt to avoid repeating that mistake. Given the speed at which AI capabilities are proliferating globally, the window to get this policy right is narrow. Whether you are a practitioner, a leader, or a policymaker, the practical implication is the same: we cannot afford to let blunt-instrument export controls become the default governance model for AI in cybersecurity.