White House to host AI companies to review new model-testing framework
Published: Aug 3, 2026 — CNBC
The White House will meet leading AI companies Tuesday (Aug 4) to discuss a newly completed voluntary framework for reviewing cybersecurity capabilities of advanced AI models. A White House official confirmed to CNBC.
Framework details (reported)
- Stemmed from President Trump June executive order
- Companies may give government early access to certain frontier models for up to ~30 days before public release / trusted partners
- Program cannot be used to create mandatory licensing, permitting, or preclearance
- Benchmarks and thresholds for “covered frontier models” expected to remain classified
- Completed framework not publicly released; metrics unpublished
Attendance (reported)
- Anthropic expected; OpenAI and Google expected (The Information); Meta invited per other outlets
- Administration working with broader industry partners