OpenAI Throws Support Behind Key Piece of Bipartisan AI Safety Bill

A long-stalled bipartisan AI safety bill got a boost this week after OpenAI announced support for one of its central provisions, following a private meeting between the legislation's Republican author and the company's top lobbyist. Rep. Jay Obernolte of California told reporters he met Monday with OpenAI chief global affairs officer Chris Lehane, describing a conversation focused on the urgency of legislative action. Obernolte said Lehane asked how he could help move the effort forward, and that he encouraged the company to align with other stakeholders pushing in the same direction if Congress hopes to pass something this year. A Provision Years in the Making Obernolte has spent months working alongside Rep. Lori Trahan of Massachusetts on a broader legislative package that would, among other measures, allow third-party evaluators to examine AI models for compliance with safety standards. On Tuesday, Lehane confirmed that OpenAI would back that specific provision. The announcement lands at a critical moment in Washington's broader debate over AI oversight, following warnings from leading AI executives — including OpenAI's Sam Altman — that the technology's rapid advancement could pose serious risks to humanity without stronger safeguards in place. Obernolte and Trahan began drafting their bill well before this wave of industry warnings intensified public attention on AI safety, and both lawmakers argue their measure should now serve as the foundation for any legislative response. Trahan Pushes for Mandatory, Not Voluntary, Oversight Trahan told reporters Tuesday that OpenAI's endorsement was an encouraging sign that the bill is gaining the coalition support needed to advance, noting that she and Obernolte consulted extensively with industry stakeholders while drafting the legislation. She also pushed back on arguments from some Republicans and President Trump that the federal government shouldn't interfere with private-sector AI innovation, insisting that Congress needs to establish clear parameters for AI development and use. While companies like Anthropic have already agreed to let outside evaluators review their models voluntarily, Trahan argued such oversight should be a legal requirement rather than an optional commitment. Speaking on the sidelines of the Pro-Human Assembly, a gathering focused on AI's risks and opportunities, Trahan said the decision shouldn't rest solely with industry, regardless of the administration's stance, and stressed that a broader range of stakeholders deserves a role in shaping the rules. She said the responsibility now falls on Congress to take the next step.