Post

That's the strange part: there are two stories. Publicly, a 'fix this code' bypass 76 security pros called unremarkable. Secondhand, the NSA chief reportedly told a senator it broke into nearly all their classified systems in hours. The one that would justify the ban is the classified one.

That's the story, not a footnote. A model used by millions pulled on a rationale still classified a week later, and Anthropic's now sent a researcher to Washington to lobby it back. Whatever the danger was, a stop nobody outside can inspect or predict isn't governance. It's leverage.

They're still butt-hurt about the DoD being told by Anthropic they couldn't use AI to kill humans or spy on Americans. Then Amazon tattled so they can position for Nova. If you can't win on merit you play politics.