Skip to content

unveiled Claude Mythos Preview

Anthropic just unveiled Claude Mythos Preview, and this thing looks like one of the most dangerous AI models we have seen yet. According to Anthropic, Mythos is strong enough at cyber offense that releasing it broadly right now would be reckless. It reportedly found major vulnerabilities across operating systems, browsers, and critical software, including bugs that had survived for decades. So this is bigger than another benchmark win 

 
,