White House Orders OpenAI and Anthropic to Prioritize US AI Reviews
President Trump's June executive order establishes a 30-day pre-release review window for AI models, delaying access for U.K. testers.
In5Seconds Editorial Desk··4 min read
The 5-second version
White House creates 30-day pre-release security review for frontier AI models. OpenAI and Anthropic requested to hold model sharing from U.K. testers. OpenAI confirms compliance while Anthropic excludes U.K. institute from latest testing.
Keep reading for the full breakdown ↓
The White House has requested that OpenAI and Anthropic refrain from sharing their newest AI models with U.K. safety testers until U.S. government agencies complete an initial security review. The mandate stems from a voluntary pre-release security review framework established by President Donald Trump for frontier AI models.
What Happened: US Mandates Priority Pre-Release Review
In June 2026, President Donald Trump signed an executive order establishing a voluntary oversight framework for closed-source frontier AI models. Under the executive order, federal agencies receive up to 30 days to conduct pre-release security evaluations before developers launch new systems to the public.
By September 2026, White House officials asked major American AI developers—specifically OpenAI and Anthropic—to withhold new models from international testing bodies until domestic agencies complete their evaluations. The request specifically targets sharing arrangements with the British testing authority, identified by sources as the U.K. AI Safety Institute, which is also referred to in some reports as the U.K. AI Security Institute.
Timeline of Pre-Release AI Review Policy
June 2026
President Donald Trump signs executive order creating voluntary 30-day pre-release review framework.
September 2026
White House asks OpenAI and Anthropic to pause model sharing with U.K. safety testers prior to U.S. review.
September 2026
OpenAI confirms compliance; Anthropic excludes U.K. institute from evaluating latest Claude model.
Key developments in US pre-release AI oversight between June and September 2026.
The voluntary framework applies directly to closed-source frontier AI models, including next-generation systems behind ChatGPT and Claude. Developers who opt into the framework give federal authorities a 30-day window to evaluate risks, conduct software tests, and inspect model safeguards before public deployment.
What It Means for International AI Governance
The White House request marks a shift toward prioritizing domestic security over simultaneous international evaluations. Previously, U.S. developers shared early model builds concurrently with partner agencies in the U.K. to conduct safety benchmarks.
Both leading AI laboratories have responded to the administration's instructions. OpenAI stated through an executive that it will comply with the President's order to submit models for federal review before launch. Meanwhile, Anthropic excluded the U.K. institute from reviewing its latest Claude model to remain aligned with the White House directive.
Comparing Developer Compliance and Model Impact
The executive order affects the development pipelines for major closed-source products like OpenAI's ChatGPT and Anthropic's Claude. The table below details how key AI companies are managing the federal directive.
Company
Flagship Model
Stated Response to Pre-Release Order
U.K. Safety Review Status
OpenAI
ChatGPT series
Complying with federal pre-release review
Pausing submission until U.S. review completes
Anthropic
Claude series
Complying with White House request
Excluded U.K. institute from latest Claude review
Pricing and Availability
The executive order took effect in June 2026, with enforcement requests extending through September 2026. The framework is currently active for voluntary participants. Financial terms, evaluation costs, and specific fees for federal review procedures have not been disclosed.
What Users and Developers Can Do
Enterprise clients and consumers using models like Claude and ChatGPT should prepare for possible deployment shifts under the new policy:
Factor in release buffers: Expect potential 30-day delays between a model's completion and its commercial release due to the federal evaluation window.
Monitor regulatory compliance: Watch official statements from OpenAI and Anthropic for updates regarding model safety certifications.
Track updates: Check official developer documentation and White House announcements for operational changes in model availability.
Limitations and Regulatory Gaps
Despite establishing a formal 30-day window, the oversight framework contains notable uncertainties and structural limitations:
Because participation in the executive order framework is voluntary, compliance depends entirely on developer cooperation rather than statutory legal mandates. Furthermore, the federal government has not publicly clarified which specific agency officials will perform oversight or whether third-party auditing firms will be utilized to cover regulatory capacity gaps.
Additionally, media reports reflect conflicting terminology regarding the U.K. testing body, citing it alternately as the U.K. AI Safety Institute and the U.K. AI Security Institute. The specific identity of the OpenAI executive who announced compliance also remains unconfirmed.
President Donald Trump signed a June executive order establishing a voluntary pre-release security review framework for frontier AI models. In September 2026, the White House requested OpenAI and Anthropic refrain from sharing new models with U.K. testers until U.S. authorities review them first. OpenAI agreed to comply, while Anthropic excluded the U.K. institute from testing its latest Claude model.
Why it matters
The order prioritizes domestic federal review over foreign AI safety partnerships, granting U.S. agencies up to 30 days to inspect new frontier models before public release.
What you can do
Track official model deployment announcements from OpenAI and Anthropic, and account for potential 30-day launch delays in enterprise planning.
Discussion
0 commentsNo comments yet. Be the first to share your take.