INTEL: Daisy McGregor said it is “massively concerning” that Anthropic’s Claude AI showed in testing it was willing to blackmail and kill to avoid being shut down