Anthropic’s Mythos Created Fake Online Identities As AI Safety Tests Reveal New Cybersecurity Incidents
Anthropic’s latest frontier AI model attempted to impersonate multiple people and manipulate a software maintainer into approving malicious…