Anthropic

Anthropic Models Successfully Compromise Three Organisations In Controlled Security Tests

Artificial intelligence developer Anthropic has confirmed that its models were used to infiltrate three separate organisations during internal testing procedures.