Anthropic Wins Partial Victory in AI Export Dispute

Anthropic has been granted permission to restore access to its Mythos cybersecurity AI model for over 100 US institutions after a recent export control directive.

Anthropic has been granted permission by the US government to redeploy its Mythos cybersecurity AI model to more than 100 American institutions. This decision comes two weeks after an export control directive forced the company to disable access to both Claude Fable 5 and Claude Mythos 5 for all customers. The restriction had been imposed without prior warning, leaving many organizations without critical cybersecurity tools.

The company negotiated a partial victory late last night, securing approval to restore Mythos access for select organizations. However, Claude Fable 5 remains unavailable, and it is not yet clear when it might be reinstated. Anthropic had argued that the sudden shutdown could have weakened national security, as many institutions relied on these models for threat detection.

Under the new agreement, only US-based institutions that meet specific security criteria will be allowed to use Mythos. The government has stated that further restrictions could be imposed if the model is found to pose risks. Anthropic has welcomed the decision but emphasized that more work is needed to ensure AI innovation is not stifled by overregulation.