Anthropic was founded in 2021 to ensure humanity can control AI and benefit from it in the long term. As the frontier lab approaches its initial public offering, though, there is one group being shut out of that control: anyone who buys its stock on public markets. Silicon Valley has long seen complex structures that let founders like Elon Musk or Mark Zuckerberg control public companies without owning most of them, but Anthropic’s AI-safety mission has produced what TechCrunch describes as a truly baroque governance system. Common shareholders will have little say over most major decisions, including who sits on the board.

The company is a public benefit company, a for-profit corporation that legally allows it to weigh a public mission alongside profit. That status protects the board’s ability to make AI-safety decisions that are not purely about shareholder value without being sued by unhappy investors. The board is currently chaired by Anthropic president Daniela Amodei and includes CEO Dario Amodei, Netflix cofounder Reed Hastings, former Microsoft and GM CFO Christopher Liddell, Novartis CEO Vas Narasimhan, and Spark Capital partner Yasmin Razavi. There is one vacancy.

Anthropic has allocated significant control of its board to the Long Term Benefit Trust (LTBT), an independent panel meant to keep the company focused on its mission to benefit humanity. The trust owns a class of stock that lets it elect and remove a majority of the board, known as Class T directors. The LTBT is chaired by Neil Shah, a global health expert and former managing director at GiveWell, and includes former Federal Reserve chair Ben Bernanke, national security expert Richard Fontaine, and Harvard physicist Christopher Stubbs. The trust has one vacancy, and existing members will choose who fills it.

Third, Anthropic’s seven founders jointly own a special class of stock. If a majority of them agree, they control 50.1% of shareholder voting power and can elect three board members. If there is no founder majority, their shares are voted in proportion with common stock. The board can also check the LTBT: when the trust nominates one of its ‘Specified’ directors, or removes that director without cause, the board’s nominating committee can decline to approve the move and send the decision to employees for a vote.

The dynamics of such a complex structure are hard to predict, TechCrunch notes, and a similar structure was essentially overthrown in OpenAI’s ‘blip,’ when OpenAI’s board briefly fired Sam Altman over concerns about candor and then reinstated him after employee loyalty forced a reversal. Anthropic’s structure seems to formalize that kind of employee power. Cracks in the founders’ united front could give common shareholders more say than expected, while employees could overrule the LTBT on board seats if they side with the board over the assembled luminaries.

Founder control does not last forever. It ends ten years after the IPO, or automatically when fewer than three founders remain. After that, the LTBT continues until 85% of common shareholders vote to end it, or it steps down with board approval. Only when neither special class exists—hopefully not because of an AI-driven catastrophe—do common shareholders finally gain ordinary control.