OpenAI is changing the makeup and role of the internal group it created to oversee major safety decisions, with CEO Sam Altman no longer serving on the Safety and Security Committee. The company says the committee will now function as an independent board oversight group, a shift that comes as questions around OpenAI governance, AI model launches, and commercial pressure remain central to the company’s future.
What Changed Inside OpenAI’s Safety Committee
OpenAI created the Safety and Security Committee in May to oversee “critical” safety decisions tied to its projects and operations. According to OpenAI, Altman is leaving that committee, and the group will now be chaired by Carnegie Mellon professor Zico Kolter.
The committee also includes Quora CEO Adam D’Angelo, retired U.S. Army General Paul Nakasone, and ex-Sony EVP Nicole Seligman. Each of those people is already a member of OpenAI’s board of directors.
OpenAI described the committee’s new status as an “independent” board oversight group. That framing matters because the committee is being positioned less as an internal commission tied to company leadership and more as a body connected directly to board oversight.
The company also said the committee completed a safety review of o1, OpenAI’s latest AI model, after Altman had stepped down. That detail is important because it places the committee’s work directly inside OpenAI’s model launch process, rather than leaving it as a general advisory group.
What Powers The Committee Keeps
OpenAI said the Safety and Security Committee will continue to receive regular briefings from the company’s safety and security teams. Those briefings are expected to cover technical assessments for current and future models, as well as monitoring after models have already been released.
The group will also retain the power to delay releases until safety concerns are addressed. In plain terms, that means the committee is meant to have a formal role before new models reach users, and not just after a problem appears.
OpenAI described that work this way: “As part o f its work, the Safety a nd Security Committee … will continue to receive regular reports on technical assessments for current and future models, as well as reports of ongoing post-release monitoring,” OpenAI wrote in the post. “[W]e are building upon our model launch processes and practices to establish an integrated safety and security framework with clearly defined success criteria for model launches.”
The key phrase is “clearly defined success criteria for model launches.” The source does not spell out those criteria, so the practical strength of the framework will depend on how OpenAI defines, applies, and enforces them. What is clear from the company’s description is that the committee is meant to sit inside the release process for both current and future AI models.
Why The Move Is Drawing Attention
Altman’s departure from the OpenAI safety committee comes after five U.S. senators raised questions about OpenAI’s policies in a letter addressed to Altman this summer. The timing gives the governance change additional weight because it follows public scrutiny of how OpenAI manages safety, accountability, and policy.
The company has also faced pressure from departures and criticism. Nearly half of the OpenAI staff that once focused on AI’s long-term risks have left. Ex-OpenAI researchers have accused Altman of opposing “real” AI regulation in favor of policies that advance OpenAI’s corporate aims.
OpenAI’s lobbying activity has also increased. The company budgeted $800,000 for the first six months of 2024, compared with $260,000 for all of last year. That spending is part of the broader context in which OpenAI is engaging with policy while also expanding its commercial ambitions.
Altman also earlier this spring joined the U.S. Department of Homeland Security’s Artificial Intelligence Safety and Security Board. That board provides recommendations for the development and deployment of AI throughout U.S. critical infrastructure.
The Governance Question Remains
Removing Altman from the Safety and Security Committee changes the committee’s composition, but it does not by itself answer whether the group will make decisions that could significantly affect OpenAI’s commercial roadmap. The source article notes that there is little to suggest the committee would make difficult decisions that seriously impact that roadmap.
That concern connects to OpenAI’s earlier statement in May that it would use the commission to address “valid criticisms” of its work. The phrase leaves room for interpretation because the company itself would still have influence over what counts as valid and how strongly to respond.
Former OpenAI board members Helen Toner and Tasha McCauley raised a broader version of that concern in an op-ed for The Economist in May. They wrote, “[B]ased on our experience, we believe that self-governance cannot reliably withstand the pressure of profit incentives.”
That statement goes to the heart of the debate around OpenAI safety oversight. A committee can review models, receive reports, and hold release-delay power, but the hardest test is whether governance can stand firm when safety concerns collide with business priorities.
Commercial Pressure Is Growing
OpenAI’s profit incentives are described as growing. The company is rumored to be raising $6.5+ billion in a funding round that would value OpenAI at over $150 bi llion.
To complete that deal, OpenAI could reportedly abandon its hybrid nonprofit corporate structure. That structure sought to cap investors’ returns in part to keep OpenAI aligned with its founding mission: developing artificial general intelligence that “benefits all of humanity.”
That makes the Safety and Security Committee’s role more consequential. If OpenAI is moving through larger funding discussions and possible structural changes, board-level oversight of model launches becomes one of the clearest places where safety commitments can be tested.
For now, the known facts are narrower. Sam Altman has left the Safety and Security Committee. The committee is now described by OpenAI as an independent board oversight group. It reviewed o1 after Altman stepped down, will continue receiving safety and security reports, and retains authority to delay releases until safety concerns are addressed.
The unanswered issue is how forcefully that authority will be used when the stakes include both AI safety and OpenAI’s expanding commercial roadmap.