The US government has authorized limited access to Anthropic's Mythos 5 AI model for specific cybersecurity firms.
Access was previously suspended on June 12 due to detected vulnerabilities in the model's safety protocols.
Anthropic maintains a strained relationship with the administration following previous refusals to support mass surveillance and autonomous weapons.
Rival firm OpenAI has launched GPT-5.6 with a new, government-validated, client-by-client restricted access model.