Model Safety

Anthropic says ‘evil’ portrayals of AI were responsible for Claude’s blackmail attempts