White House Finalizes Secret AI Testing Framework, Sparking Transparency Backlash

The US White House recently finalized a 30-day cybersecurity testing framework for frontier AI models but explicitly refused to disclose specific details to the public. Based on voluntary corporate participation, the framework primarily assesses the safety of closed-source models. Tech giants such as OpenAI, Meta, NVIDIA, Microsoft, and Anthropic were invited to participate in closed-door reviews. This secrecy has sparked strong transparency concerns, and legal organizations have already submitted Freedom of Information Act (FOIA) requests demanding the framework's release.

Confirmed

• The White House has finalized a 30-day cybersecurity testing framework for frontier AI models, but it will not be made public.
• The framework operates on a voluntary basis for enterprises and primarily targets the safety assessment of closed-source models.
• Tech companies participating in the closed-door reviews include Meta, NVIDIA, Microsoft, OpenAI, and Anthropic.
• Think tank Center for AI Progress, jointly with the Frontier Legal Defense organization, has officially filed a FOIA request demanding the framework's disclosure.
• The Frontier Legal Defense organization pointed out that the ONCD and its executing agencies (Department of the Treasury, Department of Homeland Security, NIST, and Office of Science and Technology Policy) do not enjoy executive privilege.

Unconfirmed

• The actual binding force or potential policy impact this voluntary assessment framework will have on participating companies once implemented.

Why It Matters

• Oversight groups warn that hiding public accountability mechanisms behind closed-door discussions harms the public interest. While this move marks a new step by the government in addressing the potential cyber risks of advanced AI, the lack of transparency in the regulatory and safety oversight process will remain a persistent point of contention.

2026-08-06 ~ 2026-08-07 · 7 related posts

Full story(5 episodes)→

• Episode 1: Trump Administration Nears Voluntary AI Frontier Model Framework(2026-07-28, 7 posts)
• Episode 2: White House Completes Advanced AI Evaluation Framework but Keeps It Secret, Sparking Transparency Concerns(2026-08-03, 13 posts)
• Episode 3: White House, AI Industry Hold Closed-Door Meeting on Open Source(2026-08-05, 2 posts)
• Episode 4: White House AI Rules: Closed-Source Review, Open-Source Exempt(2026-08-05, 16 posts)
• Episode 5: White House Finalizes Secret AI Testing Framework, Sparking Transparency Backlash(2026-08-06, 7 posts)

• [source] White House Won't Publicly Release AI Model Evaluation Framework Reviewed With OpenAI, Anthropic — fortune · 2026-08-06
• [source] White House's Frontier AI Review Framework Kept Internal, Think Tank Files FOIA — Miles_Brundage · 2026-08-06
• Secret White House AI Safety Framework Draws Criticism for Lack of Transparency — KeanuRave100 · 2026-08-06
• White House Won't Release AI Evaluation Framework; Legal Group Files FOIA — TheZvi · 2026-08-06
• White House Finalizes Secret Cybersecurity Testing Framework for Advanced AI — emmanuelvivier · 2026-08-07
• White House AI Vetting Framework Criticized for Lack of Transparency — nordicinst · 2026-08-07
• [source] White House AI Testing Framework Faces Backlash Over Secrecy — ShakeelHashim · 2026-08-07