OpenAI's Board Needs to Act as Altman Steps Away

Sam Altman, OpenAI’s CEO, has left the company’s internal Safety and Security Committee, which was created in May to oversee critical safety decisions. This is part of OpenAI's move to turn the committee into an “independent” board oversight group. Altman's departure and the restructure of the Safety and Security Committee come after five U.S. Senators sent an open letter to Altman, questioning him about his approach to safety and security.

Almost half of OpenAI’s staff who were working on mitigating long-term risks quit this year, and multiple ex-OpenAI staff have accused Altman of opposing AI regulation in favor of advancing corporate objectives. This should come as no surprise since the company budgeted $800,000 for federal lobbying, and the number continues to increase. For reference, the spend so far this year exceeds the budget and is four times the amount spent last year. 

So what is OpenAI's approach to safety and privacy, particularly in the context of leadership decisions and the restructuring of its Safety and Security Committee? While OpenAI positions itself as a leader in AI safety, internal and external critiques suggest that the company prioritizes rapid product development and corporate objectives over rigorous safety protocols. This situation is exacerbated by Altman's exit from the Safety and Security Committee.

Board members must take decisive action to ensure that safety and transparency are at the forefront of OpenAI's operations, given the potential impact of artificial general intelligence (AGI) on national security, ethical standards, and public trust.

Where is the transparency?

1. Internal Safety Concerns and Oversight Issues:OpenAI employees have expressed concerns over the company's safety practices, citing rushed safety testing and prioritizing product launches over thorough safety protocols.Half of the staff focused on mitigating AI risks have quit this year, citing a culture that favors corporate objectives over safety.
2. Restructuring of the Safety and Security Committee:Altman has left OpenAI's internal Safety and Security Committee, which was created in May to oversee critical safety decisions, turning it into a so-called "independent" board oversight group.The committee is now chaired by Carnegie Mellon professor Zico Kolter, but the other members are from OpenAI's board of directors, raising questions about the true independence of the oversight process.As board members are inherently tied to OpenAI’s profitability and shareholder interests OpenAI is putting their directors in a precarious situation by creating a conflict of interest.Formation of the committee mirrors Meta's formation of an Oversight Board for content policy decisions, though Meta’s Oversight Board does not include any of its fiduciary board members, further signaling a lack of independence at OpenAI.
3. External Pressure and Regulatory Resistance:OpenAI has increased its federal lobbying budget to $800,000, with actual spending exceeding this amount and being four times what it spent last year. This spending pattern suggests a strategic move to influence regulation in its favor.Former OpenAI staff accuse Altman of opposing AI regulation to advance corporate interests.
4. Public Relations vs. Safety:OpenAI has announced collaborations, such as with Los Alamos National Laboratory, to showcase its commitment to safe AI development, but critics view these as reactive measures rather than proactive safety strategies.OpenAI's website states that the revised oversight board will "receive regular reports on technical assessments for current and future models" and can delay the release of new models until safety concerns are addressed. However, whether this process will effectively prioritize safety over profitability remains uncertain.
5. National Security and Ethical Implications:Reports, including from the US State Department, highlight that advanced AI development carries national security risks comparable to those of nuclear weapons.The concentration of AGI development within a few companies like OpenAI raises ethical concerns about the societal impact of such powerful technology.

Why This Matters

The centralization of AI development within a small number of companies like OpenAI poses significant risks to national security, ethical standards, and public trust. Given the immense power and potential impact of AGI on society, the lack of rigorous and independent safety oversight could lead to destabilizing global security, ethical breaches, and misuse of AI technology. OpenAI's actions, including restructuring its safety committee to consist of internal board members, suggest a potential conflict of interest where profitability may overshadow safety considerations.

Recommendations for the Board of Directors 

1. Establish Truly Independent Oversight: Form a genuinely independent safety and ethics committee with external members who are not tied to OpenAI's board or financial interests. This group should have the authority to review and influence AI development without corporate pressure.
2. Implement Robust Safety Protocols: Develop and publicly disclose a comprehensive safety framework that includes thorough, multi-stage testing, long-term risk assessment, and external audits to ensure transparency and accountability in AI development.
3. Cultivate a Culture of Safety and Ethics: Promote an internal culture where safety and ethical concerns can be raised and addressed openly. This includes reinstating a dedicated safety team with the authority to halt or delay product releases if necessary.
4. Engage in Transparent Regulatory Dialogue: Engage constructively with regulatory bodies and participate in the creation of industry-wide safety standards. Shift the focus from lobbying for corporate interests to advocating for regulations that prioritize societal safety and ethical AI use.
5. Address National Security Concerns: Collaborate with national security experts to understand and mitigate the risks posed by AGI. Develop a crisis management plan to address potential safety breaches and the misuse of AI technology.
6. Monitor and Address Ethical and Societal Impacts: Conduct regular assessments of the ethical and societal impacts of OpenAI’s technologies. Ensure that AI development aligns with broader societal values, human rights, and ethical principles.

By taking these steps, the board can enable more transparency while protecting OpenAI’s intellectual property and assisting them in developing industry-leading AI responsibly. While innovation requires iteration, and no company is perfect, these steps can provide the building blocks, with the safety and well-being of society as the guiding principle.