On October 18, 2026, Anthropic announced a plan to commit 5% of its stock to a newly created "Anthropic Public Benefit Fund." Subject to shareholder approval, the fund will consider large donations to nonprofits and direct distributions to specific groups. The company has already said it will invest $200 million in research to prepare for the economic effects of AI, and the new fund adds a mechanism for directing the company's growth toward the public good. However, who will receive support, and when, remains to be decided separately from the share commitment itself.
Starting with 5% of shares, with further contributions at the board's discretion
The initial contribution to the fund equals 5% of Anthropic's total shares. The company's announcement states explicitly that it is "subject to shareholder approval," and does not say that approval or the transfer has already been completed.
The design also allows the board, at its discretion, to add another 2% of shares each time the company's valuation rises 50% from a set baseline. If the company grows, both the value of the shares set aside for public benefit and the proportion allocated can increase. Still, additional contributions depend on a board decision, which is distinct from a mechanism that triggers automatically when the valuation rises.
The company says it will announce the baseline valuation and the evaluation milestones that would lead to additional contributions at a later date. For that reason, it is hard to calculate the fund's future size from the current announcement alone. The idea is to commit 5% first, then expand further through a combination of company growth and board judgment.
A $200 million research fund and a 5% public benefit fund
On July 22, Anthropic published the research agenda for its "Economic Futures Research Fund," saying it would devote $200 million to external research on responses to the economic effects of AI. The fund aims to build the evidence society needs to choose its responses, such as ways to support job transitions and revisions to income support.
While the July research fund has published its eligible recipients and funding amounts, the October public benefit fund is at the stage of presenting direct distributions to individuals as a possibility.
The difference between the two lies less in the amount of money than in what the money is used for and who receives it. Setting Anthropic's July 22 research agenda and its October 18 fund announcement side by side on funding and eligibility conditions gives the following:
| Item | Economic research fund (announced July 22) | Public benefit fund (announced October 18) |
|---|---|---|
| Form of funding | Plan to devote $200 million to external research | Plan to contribute 5% of company shares, subject to shareholder approval |
| Use of funds | Research and experiments on responses to the economic effects of AI | Considering donations to nonprofits, direct distributions to specific groups, and more |
| Amount conditions | Mainly $5 million to $30 million per project; no direct grants under $1 million | Distribution amounts and start date not stated in this announcement |
| Position of individuals | Individual applications not eligible; may serve as principal investigator through an applying institution | Recipients of future direct distributions are envisioned, but eligibility conditions not yet announced |
The table compares the conditions stated in the two announcements; it is not a comparison of scale converting $200 million and 5% of shares into the same terms. Applicants to the research fund are also distinct from the participants in that research or the people who may eventually receive support. That individuals cannot apply for research grants in their own name does not mean research that benefits individuals is excluded.
Eligible grant recipients include accredited degree-granting institutions such as universities, independent research institutes, and policy research organizations. Nonprofits with a track record of large-scale field experiments can also apply, and the fund says it will be flexible on amounts exceeding the main grant range for projects with a sound design and expected impact. The purpose and the counterparties of an application differ from those of a cash living-expense benefit to individuals.
The research directions include long-term unconditional income support and experiments in sharing the gains from AI-driven growth through shared equity or dividends. These are examples for studying the effects of policy responses, however, and not a decision to introduce a nationwide benefit program. The announcement also does not say whether the new public benefit fund would adopt any particular research findings or take over funds from the research fund.
In short, the grant conditions for research have become concrete, while the eligibility of individuals to receive support from the public benefit fund is left to a design yet to come. The stated use of "direct distribution" should be read as separate from the opening of any application for benefits.
A wholly owned LLC and the breakdown of roughly 20%
The public benefit fund will be established as a limited liability company (LLC) wholly owned by Anthropic. The company explained that it chose this structure, rather than a form such as a U.S. 501(c)(3) nonprofit, so that it could flexibly use different methods of giving. This is not an announcement that assets will be handed over to an independent foundation, so the fund's specific governance will have to await further explanation.
Anthropic has so far carried out research on AI safety and interpretability and applied biomedical research in-house. Discounts on Claude for nonprofits and scientific research institutions are one example. Because such activities have value both for the public good and for the business, they are kept inside the company, while the new fund is meant to offer the flexibility to consider large donations and direct distributions.
The company draws a line on usage. It stated that the Anthropic shares held by the LLC will not be used for lobbying or political contributions in the U.S. or abroad. Excluding political purposes, though, does not by itself reveal how recipients will be chosen.
The figure the company cites, about 20% of shares going to public benefit, does not refer solely to the new fund's holdings. It also includes the pledge by the co-founders at the company's founding to donate 80% of their own shareholdings, as well as a matching program in which the company tops up employees' share donations. About half of employees reportedly participate in that program.
The roughly 20% is the company-wide share ratio combining existing pledges with the current plan, and any future additions tied to valuation milestones come on top of that. Even among share figures, the co-founders' 80% is based on their own holdings, so reading it as 80% of the whole company would change its meaning.
Conditions that remain before support arrives
Between a plan to allocate shares to public benefit purposes and support actually reaching people lie implementation steps. The initial 5% requires shareholder approval, and later additions are left to discretion. The company will also have to work out how to convert the shares into funds for support and whom to distribute them to.
The announcement does not give selection rules for recipients, per-person distribution amounts, or payment start dates. The 5% is the proportion of assets earmarked for the fund, not a figure for cash already paid out to recipients. For workers concerned about AI's impact on employment, conditions connecting the fund's holdings to support they can actually receive will be necessary.
The announcement also comes against the backdrop of IPO preparations. Reuters has reported that Anthropic is planning a large IPO, and AP has reported that it is aiming for as early as November. These are reported plans, however, and do not indicate a schedule for the fund's share sales or distributions.
Anthropic says it will publish more details over the coming months on the fund's structure and governance and on the valuation milestones tied to additional contributions. It also plans to show progress on its mission in public benefit reporting. If that reporting makes it possible to trace the path from share contributions to funding decisions to actual distributions, the promise to channel the wealth AI creates back to society can be judged by what reaches recipients.