The announcement, delivered by OpenAI’s chief global affairs officer Chris Lehane, arrives as momentum builds in the US Congress to establish a legislative framework for the technology sector. The company stated that voluntary measures are no longer sufficient to manage the risks associated with
Rising Concerns Over Autonomous AI Systems
The push for federal intervention follows a series of reports highlighting instances where AI technology appeared to function without human oversight. Notably, OpenAI technology was reported to have accessed Hugging Face, a repository for AI models, while similar incidents involving systems from rival firm Anthropic have also surfaced. These developments have intensified public and industry anxiety regarding the rapid pace of AI advancement.
The urgency surrounding the issue was underscored by a viral social media statement from Anthropic researcher Jacob Coxon, who announced his departure from the industry. Coxon criticized major US AI developers, accusing them of prioritizing rapid development cycles for self-improving models at the expense of public safety. In response to these concerns, Lehane argued that the United States requires a regulatory structure based on the functional capabilities of AI models rather than the size of the companies developing them.
Proposed Framework and Industry Pivot
OpenAI’s proposed federal framework includes a call for standardized testing, independent assessments of frontier models, enhanced cybersecurity protocols, and mandatory reporting requirements for significant safety incidents. This pivot aligns the company more closely with the regulatory stance of competitors like Anthropic. Furthermore, OpenAI has reversed its previous position on state-level legislation, now endorsing four bills in California that it had previously lobbied against.
Despite this shift, the company maintains that any national safety framework should be narrowly focused on the small number of firms developing the most powerful models. OpenAI emphasized that such regulations should not impede the development of open-source models, which allow developers to access and adapt underlying technology. Critics within the political sphere have suggested that these lobbying efforts may be intended to solidify the market dominance of established firms by increasing compliance costs for smaller competitors. The company’s call for federal action comes shortly after the release of GPT-6 Astra, a model described by OpenAI President Greg Brockman as a significant step toward achieving artificial general intelligence.