◢ AGI Lab Power Maps

San Francisco · USA · NONPROFIT-CONTROLLED PBC

Who controls OpenAI?

ULTIMATE CONTROL The OpenAI Foundation board — 8 directors incl. CEO Sam Altman

The nonprofit Foundation appoints and can remove the entire for-profit board at any time.

62/100concentration of ultimate control, the essays' editorial reading (0 = structurally checked, 100 = one person decides). How to read it

Chain of command

  1. THE TOOLOpenAI Group PBC

    For-profit public-benefit corp that builds and ships the models.

  2. CONTROLS ITOpenAI Foundation

    Nonprofit; holds special voting rights + a 26% stake (~$130B). Appoints/removes all PBC directors.

  3. RUNS THE FOUNDATION8-person board

    Bret Taylor (chair), Adam D'Angelo, Sue Desmond-Hellmann, Zico Kolter, Gen. Paul Nakasone, Adebayo Ogunlesi, Nicole Seligman, Sam Altman.

  4. SAFETY CHECKSafety & Security Committee

    Sits inside the Foundation; oversees safety across all of OpenAI.

  5. OPEN QUESTIONWho checks the board?

    The directors are self-perpetuating and answer to no electorate or external trust.

At a glance

ModelGPT series, frontier AGI mission
FormOpenAI Group PBC, controlled by OpenAI Foundation
EquityFoundation 26% · Microsoft 27% · others 47%
Workforce~4,500 → 8,000 staff [10]

Ultimate control: OpenAI Foundation board (incl. Sam Altman). Foundation appoints & can remove the entire PBC board at will. No employee vote. [1]

The pitch and the record

They say: “We're a capped-profit nonprofit. No one captures the upside of AGI.”

The cap came off. In 2025 OpenAI's for-profit became a public benefit corporation with no ceiling on investor returns. Twelve former employees filed a brief saying the company abandoned its nonprofit roots. [2] [3]

They say: “If leadership ever went wrong, the board would stop them.”

OpenAI's safety-oriented board tried exactly that in November 2023. Within days the CEO was reinstated and most of the board was replaced. The lesson the industry took away was not "respect the brakes" — it was "remove them." [1]

The essay's proposed fixes

A prescription from "The People Inside", not a fact about OpenAI and not something the lab has agreed to.

RESTORE THE CAP & THE BRAKES OpenAI removed its profit cap, converted to a PBC, and—after the 2023 board episode—rebuilt a board that answers to itself. The fix is to make the nonprofit's control real and the brakes un-removable.

  1. Reinstate a hard, binding return cap. Re-impose a legally enforceable ceiling on investor returns. Without it, fiduciary pressure to extract infinite value from AGI will eventually override the public-benefit mandate.
  2. Make Foundation control irrevocable & externally accountable. The Foundation's authority over the PBC must be locked so it can't be quietly re-papered, and its board must answer to an external body, not perpetuate itself.
  3. Give the Safety & Security Committee a binding veto. Move it from oversight-in-name to a real deployment gate that the CEO and board cannot override or dissolve, with protection from retaliatory removal.
  4. Put model weights in nonprofit custody on a revocable license. The Foundation should legally own the foundational weights; the PBC operates them under license that auto-revokes if the mission is breached.

Sources the essays cite for OpenAI

  1. [1] OpenAI — "Our Structure" & the Nov 2023 board removal & reinstatement of Sam Altman. openai.com/our-structure · en.wikipedia.org/wiki/Removal_of_Sam_Altman_from_OpenAI
  2. [2] OpenAI's 2025 conversion to a Public Benefit Corporation under the OpenAI Foundation (no return cap). openai.com/index/evolving-our-structure · nbcnews.com
  3. [3] Twelve former OpenAI employees' amicus brief; Musk v. Altman (verdict May 2026). en.wikipedia.org/wiki/Musk_v._Altman · technologyreview.com
  4. [10] Approximate workforce figures: OpenAI ~4,500→8,000 (2026); Anthropic ~3,000–5,000; DeepMind ~6,000; Meta Superintelligence Labs ~3,400; xAI ~1,200. saastr.com · engadget.com · builtin.com · time.com

Read the full essays: Who Would Own God? · The People Inside.