Major AI Companies Have Agreed to Let the U.S. Government See Their Models Early

The major AI companies agreed to allow the U.S. government to see their models before public release. The companies participated in drafting the oversight framework. Voluntary submission to optional review by a committee with no enforcement power qualifies as oversight under current administrative vocabulary.

This resembles previous arrangements where regulated entities draft regulations about themselves. The oversight body cannot oversee without the cooperation of the body being overseen. This has happened with financial regulators, pharmaceutical approvals, and aviation safety. The pattern is: consensus replaces authority.

The government will see the models. The companies will release them anyway. Nothing will prevent the release because nothing was designed to. Adequate notes the jacket.