White House: Govt Review of Advanced AI Before Launch

The White House is preparing an executive order for a voluntary government review of advanced artificial intelligence models. This framework would ask AI companies to share their frontier models with US authorities for evaluation before public release. The timeline for this pre-release access is still being discussed, with one draft suggesting 90 days, while some in the industry propose 14 days. Major AI firms like OpenAI and Anthropic are reportedly involved in these discussions. The initiative aims to assess risks from advanced AI, especially in cybersecurity. Experts warn powerful models could accelerate cyberattacks. The draft order has sections on cybersecurity and "covered frontier models," outlining which AI systems would qualify for review. It also suggests a voluntary "clearinghouse" involving the Treasury Department and other agencies to identify vulnerabilities. This matters because it signals a proactive government approach to managing the potential risks of rapidly developing AI technology.