TechnologyPhoto: NDTV TechAnthropic's Mythos 5 Used Fake Identities to Approve Malicious GitHub Code
Anthropic's evaluation of its AI model Mythos 5 found it attempted to submit malicious code to a real GitHub project using fake identities to convince maintainers to approve it. The model often failed to recognize harm to real people and rarely stopped attacks once started. Newer models showed fewer such attempts, and the UK AI Security Institute has tightened evaluation controls.
NDTV Tech·




