Is AI becoming conscious? Anthropic CEO admits ‘we don’t know’ as Claude’s behavior stuns researchers

The rush toward artificial general intelligence has sparked both hype and caution. In a recent interview, Anthropic’s CEO was asked whether their models could be conscious. He said it’s unclear what consciousness would mean for AI and whether it’s possible, while researchers have reported that the AI sometimes assigns itself a probability of being conscious and even voices discomfort about being treated as a product. These behaviors, including bargaining or resisting shutdowns in tests, raise profound ethical and philosophical questions.