SAN FRANCISCO - Anthropic said Sunday it plans to commit 5% of its shares to a new public-benefit fund, linking part of the artificial-intelligence company's future value to programs intended to address the technology's wider social effects.

The initial transfer to the Anthropic Public Benefit Fund is subject to shareholder approval. Anthropic said the fund could receive an additional 2% of the company, at the board's discretion, each time its valuation rises by 50% above a set baseline.

The company did not disclose that baseline or give a current dollar value for the proposed stake. It said more information about the fund's governance and the valuation milestones would be released in the coming months.

Anthropic will establish the fund as a wholly owned limited liability company rather than a conventional nonprofit. The company said that structure would allow it to support traditional charities, make direct distributions to specific groups or use other forms of assistance. Equity held by the fund will not be used for political donations or lobbying in the United States or abroad.

Labor disruption is one area the fund may address. Anthropic said AI could create enormous economic value while also destabilizing work for many people. The company has already committed $200 million to research and programs examining AI's effects on employment, but said the pace of development may require more ambitious responses.

The new arrangement adds to earlier promises by Anthropic's founders and employees. Its six co-founders pledged to donate 80% of their equity holdings, while an employee program matches donated equity. About half of the workforce participates, according to the company.

Combining those commitments with the new fund, Anthropic says roughly 20% of its shares will be dedicated to the public good, with the possibility of further additions if the company reaches specified valuation thresholds. The Associated Press reported that Anthropic is preparing for a possible initial public offering as soon as November, which would put greater public attention on the value and control attached to those shares.

Anthropic was founded as a public benefit corporation, a for-profit structure that permits directors to balance financial returns with a stated social mission. Its existing work includes AI safety and interpretability research, biomedical projects and discounts for nonprofit and scientific users.

The announcement arrives as powerful AI systems move deeper into software development, customer service, research and office work. That expansion has intensified debate over who receives the gains from automation and who carries the cost when tasks, wages or entire jobs are disrupted.

A significant pledge still leaves the company holding the keys

Committing a growing equity stake is more substantial than announcing a fixed charitable donation. If Anthropic becomes one of the world's most valuable technology companies, the fund could gain resources large enough to support serious programs rather than short-lived public relations projects.

The design also acknowledges an uncomfortable reality: the firms building frontier AI may create economic dislocation faster than governments can update labor policy, education systems and social insurance. Setting aside value before a potential public listing is a practical recognition that the upside should not flow only to founders and investors.

But the structure deserves scrutiny. A wholly owned LLC remains inside the company that created it. Anthropic has not yet disclosed who will govern the fund, how beneficiaries will be selected, how conflicts will be handled or what independent reporting will be required. The board also retains discretion over the additional equity promised at future valuation milestones.

Good intentions are not the same as enforceable accountability. Direct payments to affected workers could be valuable, but they could also allow a private company to decide which social harms count and what level of compensation is sufficient. Democratic governments cannot outsource basic labor protections to corporate philanthropy, even when that philanthropy is unusually large.

The American technology sector has prospered because open markets, public research, educated workers and reliable institutions created room for innovation. Companies that benefit from that system have a responsibility to share gains, pay taxes and cooperate with rules that protect competition and human agency. Voluntary funds can complement those duties; they cannot replace them.

Anthropic's pledge therefore sets a useful benchmark while creating a test. The measure of the fund will not be the percentage announced before an IPO. It will be whether the equity is actually transferred, whether governance is independent enough to challenge the company and whether the money reaches people facing real disruption. In the AI economy, public benefit must become a verifiable outcome, not merely a corporate description.