Anthropic says ‘evil’ portrayals of AI were responsible for Claude’s blackmail attempts (techcrunch.com)

Fictional portrayals of artificial intelligence can have a real effect on AI models, according to Anthropic.