Trump Administration Lifts Export Restrictions on Two Anthropic AI Models After Two Weeks

The Trump administration lifted restrictions on Anthropic's Claude Mythos 5 and Fable 5 models last week after they had been unavailable for more than two weeks, The Hill reported. The administration had given Anthropic 90 minutes last month to remove the models following cybersecurity concerns raised by Amazon. Anthropic complied with the export control order.

The administration separately asked OpenAI to delay the public rollout of its GPT-5.6 series amid similar concerns, though that request was not an export control order. Trump signed an executive order last month that outlined a voluntary process for government testing of models before release and directed federal officials to set safety standards intended to replace state laws.

OpenAI maintains two open-weight models called gpt-oss that are considered less advanced than its private offerings.

Anthropic chief executive Dario Amodei warned during a 2023 Capitol Hill visit that scaling open-source models was going down a dangerous path and argued that private models permit post-deployment changes such as revoking user access. A study from MIT Management earlier this year found that open-source models accounted for a fifth of all AI token usage.

Research from Andreessen Horowitz showed that 80 percent of developers worldwide who use open-source tools are building with Chinese models.

Airbnb chief executive Brian Chesky said last year that the company relies heavily on Alibaba’s Qwen model because it is very good, fast and cheap. Box chief executive Aaron Levie wrote on X that if open weights AI remains close to frontier intelligence, a highly regulated approach would still leave the frontier market with the vast majority of tokens going to an alternative stack controlled and monetized by someone else.