level: business
the trump administration has eased restrictions on anthropic's mythos 5 model, allowing it to be used by more than 100 specific us government agencies and companies. this move comes two weeks after a ban forced anthropic to withdraw mythos 5 and fable 5, both cybersecurity-focused models, from the market. the ban originally prohibited non-americans from accessing the models, but the new directive extends access to non-american employees at these organizations, including anthropic's own non-us staff.
commerce secretary howard lutnick stated that appropriate safeguards are in place for trusted partners to use the model. however, the administration did not address the release of fable 5, a variant of mythos 5 that was briefly available before the ban due to claimed stronger protections. both models were pulled after security researchers reportedly bypassed their guardrails easily. anthropic acknowledged the progress, noting it is working with the government to restore access and eventually make fable 5 available for general use again.
the partial reinstatement focuses on organizations that operate and defend critical infrastructure, reflecting a cautious approach to balancing security and innovation. anthropic continues to collaborate with the government to expand access further. the situation highlights ongoing tensions between rapid ai deployment and safety concerns, especially for powerful models with cybersecurity applications.
why it matters: this decision affects how advanced ai models are deployed in sensitive sectors, influencing cybersecurity practices and regulatory frameworks for ai safety.