DeepMind founded as an independent startup to “solve intelligence and use it to solve everything else.”
Acquired by Google after declining Facebook and Musk offers. Condition of sale: an ethics board to preserve the mission and control over AGI.
Google dissolves the promised ethics board after a single meeting.
Alphabet at the apex of the conglomerate; DeepMind operates as a subset of Google rather than an external entity.
After internal criticism, Google publishes AI Principles pledging to abstain from weapons and surveillance applications.
Alphabet merges Google Brain and DeepMind into a single division: Google DeepMind.
Further consolidation of Google’s research and DeepMind divisions.
Gemini teams moved under the single Google DeepMind umbrella.
A new ethics council is established, but staffed exclusively by Google personnel.
Google lifts the ban on weapons and surveillance applications; drops the reference to “harmful applications.”
Incorporated as OpenAI, Inc. — a non-profit, mission-driven counterweight to profit-driven Google/DeepMind. Funded by donations, tax-free, no shareholders.
Non-binding Charter commits to a “fiduciary duty to humanity” and flags race-to-the-bottom and power-concentration risks.
Pivot to a “capped-profit” structure to raise compute capital. LLCs let the non-profit arm keep formal control while the for-profit arms raise money; Microsoft partnership begins.
Microsoft invests $10B+ for a 49% economic interest in profits; non-profit retains formal control. November: board fires Altman, then reinstates him within days under investor and employee pressure.
Tender offer at a $157B valuation lets employees sell profit-participation units.
Restructured as a Delaware PBC, mission “to ensure that AGI benefits all of humanity.”
Transition complete: the non-profit retains control of the new PBC (nominates its directors) and owns 26% of stock; overall valuation ~$300B.
Founded by former OpenAI staff (incl. Dario Amodei) who left over safety and commercialization concerns. Incorporated as a Delaware public benefit corporation from the outset.
Disinterested experts take a long view on safety over rapid commercialization, but profit driven shareholders retain the ability to roll back the trust.
Mission stated as “responsibly develop and maintain advanced AI for the long-term benefit of humanity.”
Responsible Scaling Policy v2.2 marks the high-water mark of voluntary safety measures — later dialed back in v3.0, citing competitive pressure.
Made with the assistance of Claude Sonnet 4.6.