Helen Toner shares her side
A high-profile breakdown at OpenAI’s board is prompting scrutiny of how the company is governed and whether its nonprofit structure can realistically prioritize “humanity” over investors and employees expecting a big payout. Commenters debate whether the board was right to try to remove the CEO over alleged dishonesty and governance concerns, or disastrously naive about power dynamics, legal risk and staff backlash. The episode is also fueling broader skepticism about the seriousness of AI safety as a field, the influence of effective altruism, and whether any current structure can credibly oversee potentially existential AI technologies.
OpenAI’s Hybrid Structure, Fiduciary Duty, and Tax Constraints
- Several comments unpack the nonprofit–for‑profit tangle: a 501(c)(3) charity owns controlling entities in a capped‑profit operating structure.
- Converting to a normal C‑corp is said to trigger tax and charity‑law issues: charitable assets (including control stakes and IP) must stay dedicated to the mission or be donated/sold under strict rules, with “self‑dealing” heavily restricted.
- Some argue this structure will keep generating governance crises; others note similar vehicles are already used to mix philanthropy and startups.
Why the CEO Was Fired: Trust vs. Safety
- The former board member emphasizes the firing was about lack of trust, not immediate “AI safety” concerns.
- A key allegation: the CEO privately lobbied directors to remove a particular board member and misrepresented what others had said, leading to “inconsistently candid” behavior and deepened distrust.
- Many commenters think the board had the formal authority but executed the move disastrously: no clear explanation, no succession plan, massive backlash, then a rapid reversal.
Board Member’s Role, Expertise, and Effective Altruism
- Some see the ousted director as an important AI‑policy/safety voice connecting US–China and think their loss is significant.
- Others question their academic record, industry experience, and youth, calling them a policy wonk, lobbyist, or “childish idealist,” and view the appointment as driven by connections or ideology (Effective Altruism) rather than deep technical chops.
- The broader EA movement itself is debated: critics call it cult‑like; defenders point to its internal criticism and grant‑making track record.
Power Dynamics, Employees, and Investors
- Many note that the CEO’s fundraising, relationship with the main strategic partner, and an imminent secondary sale giving employees large payouts created massive leverage against the board.
- Employees’ near‑unanimous revolt is interpreted by some as loyalty to a strong leader, by others as motivated by financial incentives.
- Several argue that in practice “might makes right”: whatever the charter says about a duty to “humanity,” control flows to whoever can raise money and retain talent.
Meaning of “AI Safety” and Governance Lessons
- Multiple commenters complain that “AI safety” is vague and often masks either profit protection or regulatory capture for incumbents.
- Others insist there is a real technical and policy field here, citing existing research and courses.
- There’s wide agreement that the episode exposes deep weaknesses in OpenAI’s governance and that legal form alone cannot prevent a charismatic CEO from dominating a board.