Anthropic puts 5% of its shares into a new public benefit fund

The company behind the Claude chatbot is creating a new entity called the Anthropic Public Benefit Fund. It will own part of Anthropic and use that stake for public-benefit work.

Here are the terms, from Anthropic's post:

• Starting stake: 5% of Anthropic's shares, committed on day one.
• Growth rule: the board can add 2% more for every 50% increase in Anthropic's value above a set baseline.
• Approval: the plan still needs a yes from Anthropic's stockholders.
• Legal form: a wholly owned LLC, not a classic charity.

Reuters, which reported the news the same afternoon, described the stake as one that could rise as the company's valuation grows. Unite.AI and RuntimeWire also covered the plan within hours of the post.

Think of it like a staircase. The fund starts on the 5% step. Each time Anthropic's value climbs another 50% above a starting line, the board may add another 2% step.

The key words are "at the board's discretion." The extra steps are allowed, not automatic. Anthropic hasn't published the baseline value or the exact milestones yet. It says it will share them in the coming months.

Why it matters: Anthropic compared the structure directly to how tech companies pay their bosses. It is common, the company wrote, for tech firms to give executives "escalating compensation awards" as the company grows. Anthropic says it is sending that kind of growing equity to the public benefit instead.

In real life Picture a bakery that promises a local food bank 5 loaves out of every 100 it bakes, plus a couple more each time the bakery gets much bigger. The food bank's share grows with the business, not with a yearly budget meeting.

Most big donors use a 501(c)(3), the US tax category for nonprofits. Anthropic chose a wholly owned LLC instead. The company says that format gives it "adjustability between different forms of giving."

That flexibility shows up in the examples Anthropic gave. The fund might:

• Give to nonprofits: make large donations to traditional charities.
• Pay people directly: make "direct distributions to specific populations," in Anthropic's words.
• Try bolder ideas: fund approaches that a normal grant program might not cover.

There is one firm limit. Anthropic says none of the Anthropic equity held by the LLC will be used for lobbying or political giving, in the US or abroad.

RuntimeWire points out that the real impact depends on two things that aren't settled yet: the stockholder vote and the fund's governance rules, which Anthropic hasn't published.

Anthropic named one example up front: jobs. The company wrote that AI "may generate enormous economic value, but it could also cause instability for many workers."

It already has money on that problem. Anthropic says it has committed $200 million to studying and addressing AI's effect on the labor market. Unite.AI reports that this money backs the Anthropic Economic Futures Research Fund, whose research agenda the company published on July 22, 2026.

According to Unite.AI, that research fund:

• Project size: mostly backs projects of $5 million to $30 million, and nothing under $1 million.
• Who can apply: universities, independent research institutes, policy groups and nonprofits that run large field experiments, not individuals on their own.
• Focus areas: five priorities, from shaping how AI changes work inside companies to updating income support for people displaced by AI.
• Worker stakes: one priority is giving workers a share in AI-driven growth before any disruption arrives.
• Reach: it is global in scope, though Anthropic notes its funding directions lean toward the US.

The research fund itself grew out of an Economic Policy Framework that Anthropic published in June 2026, Unite.AI reports. That framework laid out programs and policies for a range of possible scenarios. The stated goal is to find out which interventions actually work, and to spread the gains of AI widely while giving workers, companies and governments time to adapt.

Anthropic's view is that research grants may not be enough. "Given the pace of AI progress, bolder approaches may be needed," the company wrote, and the new fund will have "the versatility to pursue them."

Yes, and that's the twist. Anthropic is a public benefit corporation, a type of company that can legally weigh a public purpose alongside profit. It says most of its mission work happens inside the company.

Anthropic lists three examples:

• Safety research: work on AI safety and interpretability, which means figuring out what is going on inside a model.
• Biomedical research: applied research in medicine and biology.
• Discounts: cheaper Claude access for nonprofits and scientific institutions.

Those projects help the public and also have business value, so Anthropic keeps them in-house. The new fund is for things that don't fit that model, like handing money to outside groups.

The 5% isn't the first slice of Anthropic set aside for good causes. It sits on top of two earlier commitments, according to the company:

• Co-founders: when Anthropic was founded, its co-founders pledged to donate 80% of their equity.
• Employees: a donation match program lets staff pledge equity, which Anthropic then matches. About half of employees take part.
• The new fund: 5% at the start, with room to grow.

Add it up, and Anthropic says roughly 20% of its shares are now dedicated to the public good. That number can rise if the company hits the valuation milestones it has yet to publish.

Anthropic didn't tie the timing to anything specific. But the company is in a big growth stretch. Its IPO prospectus, filed in late September, showed revenue jumping from $400 million in 2024 to $4.6 billion in 2025, as we covered when the filing came out.

The big picture: a stake that is tied to the company's value matters more the bigger that value gets. A fixed donation can shrink in importance as a company grows. This one is built to keep pace.

What's next: the stockholder vote, then details on the fund's structure, governance and valuation milestones. Anthropic also says it will report on its progress against its mission in an upcoming public benefit report. You can follow it all on our Anthropic topic page.

• Claude doesn't change: this is about who owns part of the company, not about features, plans or prices.
• Watch the jobs angle: the first example Anthropic gave is help for workers affected by AI, possibly including direct payments.
• No lobbying money: Anthropic says the fund's equity can't be spent on lobbying or political donations anywhere.
• Details are still coming: who runs the fund and when the stake grows are not public yet.