The industry's leading AI models will resort to blackmail at an astonishing rate when forced to choose between being shut down or taking extreme action, according to an alarming new report from researchers at the AI startup Anthropic. The work, published last week, illustrates the industry's struggles to align their AI models, especially as it makes the leap to creating so-called AI "agents" that can operate autonomously; in most cases, that looks like being able to use a computer's desktop to do tasks such as browse the web, shop online, and read emails.  Strikingly, pretty much all of the big […]