OpenAI has published a policy framework urging both state and federal governments to take action on artificial intelligence safety. The company argues that clear rules and oversight are needed to manage the risks of advanced AI systems while still encouraging innovation.

The proposal comes as lawmakers in the U.S. debate how to regulate AI. OpenAI says a patchwork of state laws could create confusion and slow down progress, so it wants a unified federal approach. At the same time, the company acknowledges that states can act as laboratories for testing new ideas.

Key Recommendations

  • Create a federal AI agency with enforcement powers.
  • Require safety testing for the most powerful AI models before release.
  • Mandate transparency reports for AI developers.
  • Fund research into AI alignment and safety.

OpenAI also stresses the need for international coordination, warning that no single country can solve AI safety alone. The company points to its own internal safety measures, such as red-teaming and external audits, as examples of what industry-wide standards could look like.

"We believe that advanced AI can bring enormous benefits, but only if it is developed responsibly. That means governments, companies, and researchers must work together to set the rules of the road." — OpenAI statement

The document is part of a broader push by tech companies to shape AI regulation before it is written. Critics argue that industry-led proposals may be too lenient, while supporters say they provide a realistic starting point for policy.

Source: openai.com