OpenAI to Join U.S. 30-Day AI Pre-Release Review

OpenAI confirmed it will participate in the voluntary U.S. framework that asks AI developers to provide federal agencies access to advanced models 30 days before public release. The company said the arrangement will allow government reviewers to assess potential risks prior to product launches.

The executive order that established the framework was signed on Tuesday. Under the process, participating firms will submit models to a benchmarking and review routine intended to evaluate capabilities, including cyber-related risks, and to help determine when a system should be designated a “covered frontier model.” The 30-day window is intended to give federal agencies time to review submissions before public deployment.

George Osborne, OpenAI’s head of countries, made the remarks on the sidelines of SXSW in London and confirmed the company will sign up to the program. He said democratic governments should have a strong role in how the technology is used and deployed, and that OpenAI does not wait to be asked on matters of safety and security.

Osborne said the company has been engaging with policymakers and regulators beyond the United States and has offered concrete proposals for monitoring safety and security issues internationally. “We proactively suggested ways that governments can keep a track on safety and security issues, not just in the US, but more broadly,” he added.

On regulatory design, Osborne urged institutions that combine clear authority with operational flexibility so they can adapt as technology changes. He recommended regulators be powerful but able to evolve in their procedures and oversight approaches.

The announcement follows a federal jury verdict in Oakland, California, that unanimously rejected claims by Elon Musk that OpenAI abandoned its founding nonprofit mission and improperly benefited from a shift toward a for-profit structure. Musk had sought roughly $150 billion in damages. The jury returned its decision after less than two hours of deliberation.

OpenAI’s participation in the voluntary review framework commits the company to providing pre-release access under the terms set by the order and to continuing engagement with governments on safety and regulation.