“AI companies should be required to publicly disclose their internal capabilities, future projections, and any concerning warning signs observed in their models.”