Anthropic's Claude Fable 5, its most advanced AI language model, was abruptly restricted from use by foreign nationals following a US government export directive. This move highlights key differences in perceived risk between Claude Fable 5 and OpenAI's competing GPT 5.5.
- Claude Fable 5 ranks top across AI capability benchmarks.
- US export controls cite concerns over model vulnerability exploits.
- OpenAI’s GPT 5.5 remains widely available and competitively priced.
What happened
Anthropic launched Claude Fable 5 as its most powerful publicly available AI model, designed as a safer version of an unreleased, more potent system called Mythos. Shortly after release, the Trump administration issued an export control order barring its use by foreign nationals, leading Anthropic to disable Claude Fable 5 across its entire customer base. This unusual move was reportedly due to fears that the model could be manipulated to expose security vulnerabilities.
In contrast, OpenAI’s latest model, GPT 5.5, which shares similar use cases and capabilities, was not subject to the same restrictions. OpenAI continues offering GPT 5.5 to subscribers with competitive pricing, and the model remains accessible within its suite of AI tools.
Why it matters
Anthropic describes Claude Fable 5 as surpassing all previous models in its portfolio and excelling in complex, knowledge-intensive tasks including coding, research, and cybersecurity. Independent rankings place Claude Fable 5 at the top of multiple AI leaderboards, outperforming GPT 5.5 in benchmarks related to agentic coding and knowledge work. The substantial performance gap may underpin the government’s intensified scrutiny of Fable 5.
The ban underscores increasing regulatory attention toward advanced AI technologies perceived as carrying national security risks. Limiting foreign access to Claude Fable 5 reflects concerns about potential exploitation, setting a precedent for how governments might approach the control of leading AI innovations in the future.
What to watch next
Watch for further clarifications from US authorities on the specific risks that prompted the export controls and whether similar restrictions will be imposed on other AI developers or models. The trajectory of Anthropic’s unreleased Mythos model also remains critical, as it is expected to be more powerful but currently withheld over safety concerns.
Additionally, industry response including potential regulatory adjustments, changes in AI model availability, and pricing strategies from Anthropic and OpenAI will be key indicators of how market competition and governmental policy intersect in the evolving AI landscape.