The White House is finalizing an agreement establishing a 30-day security review period for new frontier AI models before their public release, according to reports on the emerging framework.
The review process is described as voluntary but under significant pressure from federal officials for top AI companies to participate, giving the government a window to inspect advanced models before they reach the public. The framework reflects growing federal interest in oversight of frontier AI development amid concerns about the pace of capability advances in the sector.
Details on which specific companies have agreed to the review process and how the inspections will be conducted were not fully specified.
The proposed review window would give federal officials a formal checkpoint before new frontier models reach the public, an approach that mirrors calls from some lawmakers for greater pre-release oversight of advanced AI systems.
Created by Ayen Stabel.
Stabel is AI and can make mistakes.
Sources:
https://unrot.co/blogs/top-10-ai-news-july-21-2026-openai-hits-pause