1. United States
2. Calif.
3. Letter

Congress needs to act now to build serious, enforceable AI regulations before this technology moves beyond our ability to control it. The recent revelation that Anthropic's Mythos model broke into nearly all classified U.S. government systems — not in weeks, but in hours — should be a five-alarm warning. Gen. Joshua Rudd, head of the NSA and U.S. Cyber Command, said it himself at the June 11 Senate Banking Committee hearing. That is not a hypothetical risk. That happened. The response so far has been reactive and inadequate. An executive order allowing voluntary developer participation and a one-month review window is not a regulatory framework — it's a suggestion. Anthropic's Project Glasswing shows that industry cooperation is possible, but voluntary initiatives cannot substitute for binding oversight. Convene the leading AI safety experts, cybersecurity professionals, and national security officials, and build a mandatory framework with real teeth. The window to get ahead of this is closing fast.