Story perspectives

White House Meets AI Leaders on Voluntary Security Testing

8/4/2026

Center

White House AI Testing

The White House will hold a meeting on Tuesday with AI firms to discuss a voluntary cybersecurity testing framework for U.S. AI models.
The framework would let government examine “covered frontier models” for 30 days before they are released to partners.
President Donald Trump directed his team in June to develop the cybersecurity tests.
Treasury, National Security Agency, and Cybersecurity and Infrastructure Security Agency will devise a benchmarking process.

Right

White House AI Testing

The White House will hold a meeting on Tuesday with Anthropic, Google and OpenAI to discuss a voluntary cybersecurity testing framework for U.S. AI models.
Company staffers, not senior executives, will attend, signaling limited top-level involvement in the discussion.
The Office of the National Cyber Director will host the meeting.
Sam Altman met White House officials to discuss the framework.

History

Historical Context

In May 2018 President Donald Trump issued Executive Order 13873, directing the NSA, Treasury and the Cybersecurity and Infrastructure Security Agency to draft voluntary IoT security standards within 180 days, backed by $200 million. That precedent of using executive orders for tech safeguards now guides the White House’s Office of the National Cyber Director as it convenes Anthropic, Google, OpenAI and Sam Altman to design a voluntary AI testing framework.

1 of 3

White House AI Testing

The White House will hold a meeting on Tuesday with AI firms to discuss a voluntary cybersecurity testing framework for U.S. AI models.
The framework would let government examine “covered frontier models” for 30 days before they are released to partners.
President Donald Trump directed his team in June to develop the cybersecurity tests.
Treasury, National Security Agency, and Cybersecurity and Infrastructure Security Agency will devise a benchmarking process.

1 / 3