OpenAI Agrees to Pre-Release AI Model Reviews Under Trump Executive Order

CNBC reported Friday that OpenAI will comply with President Donald Trump’s executive order requiring AI companies to submit their models for government assessment before public release. The confirmation came from OpenAI’s head of countries, George Osborne, speaking on the sidelines of SXSW London.

OpenAI Endorses the Voluntary Pre-Release Framework

The order, which Trump signed earlier this week, gives the federal government access to AI models 30 days ahead of their release. Authorities will use that window to run a benchmarking process designed to evaluate advanced capabilities and determine whether a given model qualifies as a “covered frontier model.”

Osborne told CNBC’s Arjun Kharpal that OpenAI intends to participate in this voluntary scheme. He described democratic governments as having a legitimate and significant role in shaping how powerful AI technology is used and rolled out across society.

He added that the company does not wait to be compelled before engaging with regulators. OpenAI has, according to Osborne, already proposed mechanisms through which governments can track safety and security concerns tied to frontier models, both in the United States and internationally.

Also Read: Axios on the Trump AI executive order signing

Background: Osborne Brings Policy Credentials to the Role

Osborne is not a typical tech executive. He served as the United Kingdom’s Chancellor of the Exchequer from 2010 to 2016 under Prime Minister David Cameron, making him one of the most senior political figures to move into a leadership role at a major AI company.

His comments at SXSW London reflect a broader shift at OpenAI toward proactive government engagement. The company has expanded its policy and public affairs presence significantly over the past two years as regulatory scrutiny of large language models has intensified globally.

Also Read: OpenAI’s government engagement strategy via Reuters

What the Benchmarking Order Means for the Industry

The pre-release review framework is designed to help the administration identify models that may pose risks tied to cyberattack capabilities or other national security concerns. Companies falling under the “covered frontier model” designation would face closer federal scrutiny.

Osborne suggested that regulatory bodies should be established with built-in flexibility. He argued that rigid frameworks risk becoming obsolete as the technology evolves rapidly and that smart governance must leave room to adapt.

OpenAI’s swift public endorsement puts pressure on other frontier labs to signal their own intentions around the order.

Read Next: AI Regulation Roundup: What the Major Labs Are Saying