The Trump administration's move to restrict foreign access to Anthropic's most advanced AI models, Mythos 5 and Fable 5, has deepened concern that Chinese players are gaining comparable cybersecurity capabilities and could erode the US edge in cyberwarfare.
June 2026 · Anthropic
US Export Controls Cut Off Anthropic's Most Advanced AI Models
A Commerce Department directive barred foreign governments, companies and individuals from using Claude Fable 5 and Mythos 5 — prompting Anthropic to temporarily shut off access for everyone to comply, amid fears it could hand China an edge in cyberwarfare.
Jun 13
Directive received under the Commerce Secretary's name
271
Firefox bugs found by Mozilla using Mythos Preview
~95.5%
SWE-bench Verified score
Cyber Gym benchmark — neck and neck
Cyber-capability score (%)
The top models are effectively level — a closed performance gap that fuels US concern China is catching up.
Fable 5
Safeguarded general-release version — now limited under the controls.
Mythos 5
Cyber-capable variant — restricted to trusted Glasswing partners only.
Timeline of the clampdown
Directive issued
Foreign govts, firms & nationals barred from both models.
→
Full shut-off
Anthropic cuts access for everyone to comply.
→
Partial rollback
Late June: Mythos 5 restored for trusted partners only.
Defensive upside
Partners like Mozilla used the model to surface hundreds of real bugs — a powerful defensive tool for finding zero-days and exploit chains.
Critics' warning
Sweeping controls could block allied banks and critical infrastructure from defending themselves — experts call it a "gigantic mistake."
Continue reading The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in ✓ Signed in — this article isn’t included in your current plan.Unlocking the full article…