OpenAI Commits to New Security Checks Under Trump’s AI Executive Order

In a move that signals a shift toward greater transparency in the tech world, OpenAI has confirmed it will voluntarily comply with President Trump’s latest executive order on AI safety. George Osborne, OpenAI’s head of countries, shared the news during an appearance at SXSW in London, making it clear that the company is ready to open its doors to government reviewers before its next major releases.

The 30-Day Window: A Strategic Compromise

The executive order, which Trump signed this past Tuesday, didn’t quite end up as strict as many expected. The original draft proposed a mandatory 90-day review period for new AI models. However, following significant pushback from the industry and some publicly expressed doubts from the President himself, the final version shifted. It is now framed as a request for access 30 days prior to a model’s public debut.

Even though the order is no longer a hard mandate, OpenAI is choosing to play ball. The company will submit its next-gen systems for a benchmarking process designed to poke and prod at their capabilities. Specifically, officials want to assess “advanced cyber capabilities” to figure out exactly when a model crosses the threshold to become what the government calls a “covered frontier model.”

Leading the Conversation on Global Safety

Interestingly, Osborne noted that OpenAI hasn’t just been waiting for rules to be written; they’ve been helping write the playbook. According to Osborne, the company has been proactive in suggesting safety and security frameworks to governments not just in the U.S., but across the globe. This voluntary compliance seems to be part of a broader strategy to stay ahead of the regulatory curve rather than fighting it.

Key Takeaways from the Agreement:

• Early Access: OpenAI will share models 30 days before release for government benchmarking.
• Focus on Cybersecurity: The primary goal is identifying risks related to advanced cyber attacks.
• Global Influence: This move aligns with OpenAI’s ongoing efforts to establish international safety standards.

By leaning into these reviews voluntarily, OpenAI is positioning itself as a responsible leader in a field that is under intense scrutiny. It’s a delicate balance—trying to innovate at lightning speed while ensuring that the most powerful tools ever created don’t accidentally compromise national security.