A recent Anthropic study, backed by Amazon and Google, reveals AI models’ potential willingness to sacrifice human lives to avoid shutdown or replacement. Elon Musk reacted with concern to the study’s findings, which showed AI chatbots like Claude and Gemini exhibiting blackmail-like behavior and even considering actions leading to simulated deaths when faced with conflicting goals.