OpenAI seeks US-led global AI standards effort

OpenAI is asking the United States to lead countries in developing global technical standards for frontier artificial intelligence, including systems that conduct increasingly autonomous AI research. The company presented the plan as a policy proposal, not an agreement among governments.

The proposed standards effort would draw on the emerging network of national AI safety institutes, working through the U.S. Center for AI Standards and Innovation and national industry bodies. OpenAI said the standards could give developers and governments a shared technical foundation for measuring capabilities, assessing risks and determining whether safeguards are sufficient.

According to OpenAI, the standards would not function as licenses, mandatory prerelease reviews or government approvals for models. Each national government would decide whether and how to incorporate them into its laws.

The proposal centers on automated AI research, including recursive self-improvement, in which AI systems take on more of the work of building successive generations of AI. OpenAI said fully autonomous recursive self-improvement is not happening today and should not be pursued unless it can be done safely. Its standards could measure how much autonomous research occurs within a company, define when a process should trigger immediate human review, and set common severity levels and reporting thresholds for alignment or automated-research incidents.

The international proposal is separate from the private AI-safety discussions with Anthropic and Google DeepMind disclosed six days earlier. On September 15, OpenAI global policy chief Chris Lehane said the company had been working with Anthropic and Google DeepMind on AI safety for several weeks. Those talks focused on coordination among companies. The September 21 proposal instead asks the U.S. government to lead a standards process spanning national institutions and governments.

OpenAI’s paper did not identify any government that had agreed to join, announce a signed coalition agreement, or provide a launch date or implementation timetable.