Letter: the US says Anthropic "agreed to proactively detect and address security risks" of Fable 5 and Mythos 5; a source says Anthropic developed a "safeguard"

US government move allows AI start-up to re-release Mythos and Fable models