Anthropic Restores Claude Mythos 5 Access for US Critical Infrastructure Defenders After Two-Week Government Review

Anthropic has confirmed the restoration of access to Claude Mythos 5 for a select group of US organisations that operate and defend critical national infrastructure, following two weeks of direct engagement with federal authorities that began on 12 June 2026. The company confirmed the development on its official X account, stating it is working with the US government to expand Mythos 5 availability further and to make Fable 5 available for general use once again, suggesting a phased authorisation process that mirrors how sensitive defence technologies are managed and cleared for deployment.

The access restoration applies specifically to organisations aligned with CISA's framework for critical infrastructure sectors, including energy grids, financial systems, telecommunications networks, and government systems. Claude Mythos 5 is described by Anthropic as its strongest cybersecurity model, purpose-built to support defensive cyber operations including threat detection, vulnerability triage, and incident response acceleration.

Why Access Was Suspended

Anthropic has not publicly disclosed the precise cause of the June 12 suspension. However the context is well established. The US Commerce Department imposed restrictions on Anthropic's Mythos and Fable AI models on June 12, blocking foreign national access on national security grounds, following a period of sustained Chinese AI model distillation attacks against Claude. Anthropic had disclosed to the US Senate Banking Committee that Alibaba conducted the largest known distillation attack against Claude, generating 28.8 million fraudulent exchanges through 25,000 fake accounts between April and June 2026, following earlier campaigns by DeepSeek, Moonshot AI, and MiniMax.

The suspension, while aimed at preventing foreign access, had the immediate effect of cutting off US-based defenders as well, including the NSA, which lost access to Mythos during the dispute, an ironic outcome given that NSA chief Joshua Rudd had testified to Senator Mark Warner that Mythos broke into almost all US classified systems within hours during Project Glasswing testing, demonstrating exactly the kind of capability that makes the model strategically valuable for both offensive assessment and defensive hardening.

What the Restoration Means

The structured re-authorisation confirms that the US government now treats advanced AI cybersecurity models as regulated defence assets rather than general-purpose software. The same scrutiny applied to sensitive defence technologies, staged access, government clearance, and approved organisation lists, is now the operating framework for frontier AI models with significant offensive and defensive security capability.

For GCC enterprises and government entities that have built AI infrastructure strategies around Anthropic partnerships, this development reinforces a point that has become increasingly clear through 2026: access to frontier AI models is no longer a purely commercial relationship. It is subject to national security policy, export control frameworks, and government-level authorisation decisions that can change with limited notice and significant operational impact.

The UAE's position as one of Anthropic's most significant international partners, formalised through the US-UAE AI Acceleration Partnership, has been directly affected by these restrictions. The G7 trusted partners framework discussed in June, through which allied nations could be granted structured access to frontier models, represents the most credible pathway for GCC governments to secure stable, policy-backed access to Mythos-class capabilities rather than relying on commercial availability alone.

OpenAI, Google DeepMind, Microsoft, and xAI have all agreed to provide the US government early access to new models for national security evaluations. The emerging pattern is consistent: frontier AI cybersecurity models are becoming a managed national security resource, not an open commercial product, and access will increasingly flow through government-mediated frameworks rather than standard API keys.

Security teams at critical infrastructure organisations should expect formal restoration notifications through their existing Anthropic service channels. Organisations not yet on the approved access list should monitor Anthropic's official communications for updates on expanded availability.