OpenAI has confirmed that it will voluntarily comply with a new executive order from President Trump related to AI model oversight, according to comments from its head of countries, George Osborne.

The order, signed on Tuesday, requests—but does not mandate—access to the latest AI models 30 days before their public release. It was originally proposed as a stricter rule requiring companies to submit models 90 days in advance, but was later revised following public criticism and expressed doubts from Trump himself.OpenAI will voluntarily comply with Trump’s new AI reviews order

Under the agreement, OpenAI will participate in a benchmarking process designed to evaluate the “advanced cyber capabilities of AI models” and determine when a system should be classified as a “covered frontier model.”

Speaking at SXSW in London, Osborne said OpenAI has already been suggesting frameworks for governments to monitor AI safety and security concerns, not only in the United States but also across other regions globally.