Anthropic Sets Aside 5% of Its Shares for a Public Benefit Fund: A Governance Test

Anthropic has announced that it will set aside 5% of its shares for a public benefit fund. The move brings back a familiar governance question at AI companies: how to weigh profit against the public interest. The size of the stake isn't the real question, though. What matters is who the fund answers to.

What happened?

Anthropic has pledged 5% of its shares to a fund with a public benefit mission. The initial announcement didn't say how the fund will be governed. It also didn't say where the money will go or when the shares can be turned into cash.

Why is the balance between profit and public benefit so contested at AI companies?

Since its founding, Anthropic has presented its governance structure as part of its identity. The company is incorporated in Delaware as a Public Benefit Corporation (PBC). Under that status, the board must weigh a stated public benefit goal alongside shareholder returns. There is a further layer on top: the Long-Term Benefit Trust. Its trustees have no financial stake in the company, and over time the trust gains the right to elect a majority of the board.

The weakness of PBC status is well known. "Public benefit" is broadly defined and hard to enforce. The right to sue directors over this duty usually belongs only to investors who hold more than a certain share of the company. In other words, the structure is more a formal statement of intent than a binding mechanism. The new fund puts a measurable asset where that gap used to be. An abstract commitment becomes a stake with real value on the balance sheet.

The comparison with OpenAI is useful here. OpenAI started out as a capped-profit company controlled by a nonprofit. Its leadership crisis in late 2023 showed how fragile that model was under pressure. The company then moved to a PBC structure, and the nonprofit foundation still holds a large stake in the new company. In short, the two labs are arriving at the same place from opposite directions: a commercial company with a public benefit entity alongside it.

What changes? Who will control the fund?

One variable will decide what the fund actually means: control. Will it be run by independent trustees, or will it answer to Anthropic's executives? Will the fund hold the voting rights on the shares? If it does, 5% is a small but meaningful voting bloc. If it doesn't, the arrangement amounts to little more than a long-term philanthropic pledge.

For shareholders, the picture is complicated too. The 5% stake dilutes existing investors. If major backers like Amazon and Google signed off on it, they probably see the cost as low and the reputational payoff as high. There is also a subtler contradiction. The fund's value is tied directly to Anthropic's market valuation, so the fund grows richer as the company grows. That means an entity that benefits from the company's commercial success is also expected to guard the public interest. This is where the claim of independence can start to weaken.

What's next?

The timing is hard to ignore. Reports that Anthropic is preparing for an IPO have been circulating for some time. At a public company, PBC obligations and a fund like this would appear in the prospectus as either a risk or a safeguard, and institutional investors would price that in. So the fund's structure can also be read as a way to lock in governance ahead of a listing.

What to watch in the coming months:

• The fund's charter and who its trustees will be
• Whether the voting rights on the shares stay with the fund
• Where the money goes: safety research, education or economic impact programs
• Any limits on selling the shares, and how the fund relates to the Long-Term Benefit Trust

On its own, the 5% fund isn't a real check on the company. If it has independent management and voting rights, it adds a measurable layer to the existing structure. Until the charter is published, though, it's also fair to read the move as a reputation play ahead of an IPO.