OpenAI has confirmed it will comply with Donald Trump’s executive order that asks AI companies to allow the US government to assess their models’ capabilities before they are released.
The order was signed on Tuesday and focuses on early access to advanced AI systems. It asks companies to provide access to their latest AI models 30 days before public release. However, it does not force firms to submit or hand over the models formally.
An earlier version of the policy was more stringent and would have required companies to submit AI models up to 90 days before launch. The proposal was later revised following public criticism, concerns from industry groups and objections raised by President Trump.
https://www.youtube.com/watch?v=lmPyOQLRTEU&t=2s
OpenAI has confirmed that it plans to participate in the review framework established under the order. The company will submit its next-generation AI models for a structured benchmarking process designed to assess advanced cyber capabilities.
The review process will also help determine whether a system should be classified as a “covered frontier model,” a category intended for highly advanced AI systems that may require additional oversight.
Speaking at the SXSW event in London, George Osborne said OpenAI has already proposed ways for governments to monitor AI safety and security risks. He added that these discussions are taking place not only in the United States but also as part of broader global efforts to address the challenges and opportunities presented by advanced artificial intelligence.
