The Commerce Department’s Center for AI Standards and Innovation and other government officials are quietly evaluating Anthropic’s new AI hacking capabilities.