White House Eyes 90-Day Early Notice Rule for Advanced AI Releases
The White House may soon require 90-day notice before releasing advanced AI models.
Why it matters: Legal and compliance teams in tech and infrastructure must prepare for potential new notification and evaluation mandates when deploying advanced AI. The changes could accelerate regulatory due diligence for law and policy professionals advising AI developers and critical infrastructure firms.
- Expected executive order mandates 90-day government and infrastructure notice for advanced AI releases.
- Move responds to finding 75 software vulnerabilities in a month using top AI tools, per Palo Alto Networks.
- 32 lawmakers from both parties urge the White House to act swiftly on AI cybersecurity risks.
- Proposed order mirrors FDA’s drug pre-clearance process for AI model safety reviews.
The White House is preparing an executive order to require developers of large-scale artificial intelligence models to notify the government and key infrastructure operators 90 days before public release. The rule would target advanced models that pose high safety or cybersecurity risks, according to administration officials.
- Palo Alto Networks reported finding 75 software vulnerabilities in a single month with help from advanced AI tools by Anthropic and OpenAI—a marked increase over typical rates. Rapid detection raises both defensive and compliance stakes for infrastructure providers.
- National Economic Council Director Kevin Hassett said the goal is a “clear road map” for AI releases, using a safety review model similar to the FDA’s drug approval process. This would require confirming the security of advanced AI before it is widely accessible.
- The bipartisan push includes 32 House members who caution that “disclosure, validation, patching, and deployment efforts may struggle to keep pace” with AI’s ability to uncover vulnerabilities, as emphasized in their letter to National Cyber Director Sean Cairncross.
The decision timeline has been delayed by White House debates and competing priorities, such as foreign policy events. However, legal teams advising AI enterprises or utilities should watch for new compliance infrastructure relating to government pre-notification and security evaluation steps.
As AI rapidly advances, future notification or pre-release evaluation mandates could form the backbone of federal AI safety regulation, directly impacting legal practice around risk assessment and policy compliance.
By the numbers:
- 75 vulnerabilities — Discovered in one month by Palo Alto Networks using advanced AI tools.
- 32 lawmakers — Number of bipartisan House members urging expedited federal AI safeguards.
Yes, but: Internal White House disagreements and major international summits are delaying final action on the order.
What's next: Watch for formal executive order language and compliance guidance in the coming weeks.