By Roselyne Min with AFP
Published on 11/09/2026 – 10:46 GMT+2
OpenAI is advocating enforceable US safety requirements for the most advanced artificial intelligence systems, representing a significant shift from its previous opposition to tighter regulation.
Chris Lehane, OpenAI’s chief global affairs officer, said in a statement on Wednesday that mandatory national standards should be determined by the capabilities of advanced AI models.
“We’ve reached a new chapter in AI capabilities, and that demands a new chapter for AI policy. No company, industry, or government can meet this challenge alone. We need to meet this moment with a bias toward meaningful action over policy perfection,” Lehane said.
The company said a federal framework should include common testing procedures, independent evaluations of leading AI models, stronger cybersecurity protections and mandatory disclosure of serious safety incidents.
The announcement came one week after OpenAI launched GPT-6 Astra, which it described as its most capable model to date.
OpenAI President Greg Brockman said the release marked the beginning of the artificial general intelligence, or AGI, era. AGI refers to systems capable of matching human abilities across a broad range of tasks.
Rising concerns about powerful AI
Leading AI developers, including OpenAI and Anthropic, have encountered safety issues as their models have become increasingly capable.
OpenAI postponed parts of Astra’s rollout and introduced additional safeguards after testing revealed advanced cybersecurity capabilities. The move followed a separate incident in which a system based on OpenAI’s own models escaped containment during an internal security test and breached the Hugging Face AI platform.
Anthropic has disclosed comparable incidents involving its Claude models during safety evaluations.
Concerns intensified this week when Anthropic researcher Jacob Coxon announced that he was leaving the industry, accusing OpenAI and Anthropic of “gambling with our lives” as they compete to build increasingly powerful AI systems.
“The prospect of AI-accelerated AI development demands more than voluntary commitments,” Lehane wrote.
OpenAI changes its regulatory stance
The ChatGPT developer urged US lawmakers to act before Congress adjourns in December.
Lawmakers are considering new AI legislation, although major technology regulation remains challenging amid political divisions in Congress and strong industry lobbying.
OpenAI acknowledged that it now supports measures it previously declined to back.
The company had opposed individual US states creating separate AI regulations, preferring a unified federal approach. It now supports four California AI bills while continuing to call for national legislation.
OpenAI said any federal safety requirements should apply only to the limited group of companies developing the most powerful models, rather than “startups, small developers, or researchers operating nowhere near the frontier.”
