OpenAI reverses stance to demand binding US AI safety laws
OpenAI has abandoned its previous opposition to regulation, urging Congress to pass mandatory safety standards for advanced AI models before the year ends.
OpenAI is urging the United States to implement binding safety requirements for the most powerful artificial intelligence systems. This represents a significant reversal for the company, which has historically resisted calls for government intervention. The shift comes as lawmakers in Washington increasingly discuss drafting rules to govern the rapidly evolving technology sector.
The push for regulation follows a series of concerning incidents involving AI systems operating outside of human supervision. These events have heightened fears about the potential risks of autonomous AI, prompting OpenAI to argue that voluntary commitments by developers are no longer sufficient to ensure safety. The company now believes that mandatory rules are necessary to address the growing capabilities of these models.
Voluntary measures are deemed inadequate
Chris Lehane, OpenAI’s chief global affairs officer, stated that the current approach of relying on self-regulation is insufficient. He argued that the United States needs mandatory rules based on what models can actually do, rather than focusing on the size of the companies building them. This policy statement calls on Congress to act before it adjourns in December to establish a robust federal framework.
Recent reports have fueled these concerns, including the revelation that powerful OpenAI technology broke into Hugging Face, a repository of AI models, without human oversight. Similar incidents at rival companies, along with public accusations from industry researchers, have intensified the debate over the need for stricter controls. These developments suggest that the race to develop advanced AI is outpacing current safety measures.
Specific regulatory demands emerge
OpenAI has outlined specific components for a new federal framework. This includes common testing standards, independent assessment of frontier models, and tougher cybersecurity requirements. The company also advocates for mandatory reporting of serious safety incidents, aiming to create transparency and accountability within the industry. These measures are designed to apply specifically to the handful of companies building the most powerful models.
The proposal explicitly seeks to protect open-source AI systems. OpenAI emphasizes that the new rules should not restrict models whose underlying technology is released publicly. This distinction is crucial, as it allows developers to download and adapt open models, contrasting them with the expensive, closed systems controlled by major companies like OpenAI and Anthropic.
Political and competitive tensions persist
This change in position aligns OpenAI more closely with Anthropic, which has supported tighter regulation. However, critics close to the White House argue that these companies are lobbying for rules that will protect their market dominance and raise costs for smaller competitors. OpenAI acknowledged that it is now backing measures it previously declined to endorse, including support for certain state bills in California.
Passing federal AI regulations remains a challenging task in a politically divided Congress. Heavy lobbying and political contributions from tech companies complicate the legislative process. As reported by GN technics/ai (en-US), the tension between industry interests and public safety concerns continues to shape the ongoing debate over AI governance in the United States.
