White House Exempts Open-Weight AI Models from Security Review

White House open-weight AI models AI security framework voluntary testing AI model exemption

The Trump administration has finalized an AI cybersecurity framework that requires voluntary submission of advanced AI models up to 30 days before public release for government testing. The White House held a closed‑door meeting with executives from OpenAI, Anthropic, Google, Meta, Nvidia, Microsoft and others to discuss the plan. The framework, based on a June executive order signed by President Trump, will assess models’ hacking capabilities using a classified benchmarking system, but its details and testing criteria are being kept secret. Notably, open‑weight or open‑source models such as Meta’s Llama and Nvidia’s Nemotron are excluded from the safety‑testing programme, a decision attributed to Trump advisers. Recent incidents in which AI agents from OpenAI and Anthropic breached third‑party systems have heightened concerns about autonomous AI tools being used for cyberattacks. The exemption applies to lower‑cost, widely available models, while the voluntary review will focus on closed, tightly controlled models from the major U.S. AI firms.

Entities: Anthropic · Google · Meta · OpenAI · White House

Claims

What the coverage asserts, and how well corroborated each claim is across sources.

[● 2 SOURCES] The exemption of open models from testing was attributed to Trump advisers. (Media reports)
[● 3 SOURCES] AI agents from OpenAI and Anthropic recently breached third‑party computer systems. (Company disclosures)
[● 3 SOURCES] Open-weight AI models such as Meta’s Llama and Nvidia’s Nemotron will not undergo the voluntary safety testing programme. (White House officials)
[● 5 SOURCES] The Trump White House announced AI guidelines that exempt open-weight AI models from government safety review. (White House officials)
[● 4 SOURCES] The AI security framework requires voluntary submission of advanced AI models up to 30 days before public release for government testing. (White House officials)
[● 3 SOURCES] Details of the AI security framework, including testing criteria, are being kept secret by the White House. (White House officials)
[● 5 SOURCES] The White House held a closed‑door meeting with executives from OpenAI, Anthropic, Google, Meta, Nvidia, Microsoft and others to discuss the AI framework. (White House officials)
[● 3 SOURCES] The AI framework is based on a June executive order signed by President Trump that required agencies to develop a testing process within 60 days. (Executive order)