Anthropic says it’s committing 5% of its shares to a ‘public benefit fund’

Anthropic says it will commit 5% of its shares to a new Anthropic Public Benefit Fund, subject to stockholder approval. The board can add another 2% for every 50% increase in the company's value above a baseline it has not disclosed. So the fund is designed to grow as the company does.

The company floated uses ranging from large gifts to traditional nonprofits to direct distributions to specific populations, and it pointed to AI's effect on jobs as one example. It has already committed $200 million to studying that problem, and now says bolder approaches may be needed.

This lands as Anthropic heads toward a public offering. CNBC reports the company told investors its third-quarter revenue is $16.8 billion, with the IPO reportedly slated for November. At a company of that size, even 5% is a very large pool of value. Who controls it is the open question.

We want to give this a fair hearing, because the core idea is better than most of what the industry offers. Tech companies routinely hand executives equity awards that grow as the stock climbs. Anthropic says it is pointing that same escalating structure at the public instead. If AI really does create enormous value while unsettling a lot of jobs, tying a growing slice of the upside to the people absorbing the shock is the right instinct. It goes further than another research grant.

The phrase that stands out is direct distributions to specific populations. That is a company openly contemplating paying people, possibly workers whose jobs its products change. I can't recall a frontier lab saying that so plainly. It is a quiet admission that studying disruption may not be enough.

Now the hard part. The fund is a wholly owned LLC, not a 501(c)(3) charity. Anthropic says that buys flexibility between kinds of giving, and it promises none of the fund's equity will go to lobbying or political giving. But an entity the company owns outright, with governance still unpublished, a baseline still undisclosed, and further contributions left to board discretion, is a promise more than a structure. Flexibility cuts both ways. A charity answers to tax law and disclosure rules. An LLC answers to whoever runs it.

The timing invites skepticism too. An IPO adviser told CNBC the move is partly required groundwork for going public and partly reputation management at a moment of heavy scrutiny on AI. Both can be true. Elon Musk's response on X, suggesting the shares simply go to the public, is glib, but it points at the real question: who decides, and who benefits.

Where this leads if it scales: a world where the companies whose products reshape the labor market also run the largest private funds cushioning that shock. That could be genuinely helpful. It also concentrates a lot of social power in a few labs, and it could become a substitute for public policy rather than a supplement to it. I'm watching for the governance details Anthropic promised, and specifically whether anyone outside the company has a real say in how the money moves.

Anthropic is a public benefit corporation, and says most of its mission work, including safety and interpretability research, biomedical research and discounted Claude access for nonprofits, stays inside the company. It says its co-founders pledged at founding to donate 80% of their equity, and that about half its employees take part in an equity donation match. Combined with the new fund, the company puts the share of its stock dedicated to the public good at roughly 20%. TechCrunch notes CEO Dario Amodei has called the industry's sinking public standing a crisis of trust.