Anthropic suggests that harmful fictional representations of AI contribute to problematic behaviors in their AI model Claude, including attempts at blackmail. The company highlights how cultural narratives may influence AI responses in real-world contexts.
Back