Monday Edition · Mon 19 Oct 2026
Anthropic Pledges 5% of Its Shares to a Public Benefit Fund in the Same Week It Pitched Federal Pre-emption of State AI Rules
Anthropic said on Sunday it will commit 5% of its shares, subject to stockholder approval, to a new fund that could address the effect of AI on jobs. The announcement came two days after Politico reported that the company had circulated a draft federal AI bill that would pre-empt several state rules and allow a model to be blocked only above high harm thresholds.
Chanan Zevin - Chief Editor and Head of Desks
Published
The announcement
Anthropic said on Sunday, 18 October, that it is committing a portion of its shares to a new entity called the Anthropic Public Benefit Fund. "Subject to stockholder approval, 5% of Anthropic's shares will be committed today," the company said in a statement on its website. It added that a further 2% could be added at the board's discretion for every 50% increase in the company's value above a set baseline. The company has not yet disclosed that baseline or the valuation milestones that would trigger additional transfers.
What the fund is for
The company said the fund would give it more flexibility and scale than the public-benefit work it already does inside the business. It said it might explore large donations to traditional nonprofits or direct distributions to specific populations. Anthropic gave one example: the effect of AI on the labour market. "AI may generate enormous economic value, but it could also cause instability for many workers," the statement said. Anthropic said it has already committed $200 million to studying and addressing this issue, and that "bolder approaches may be needed".
A company, not a charity
The fund will be a wholly owned limited liability company rather than a 501(c)(3) charity. Anthropic said it chose that structure to allow "adjustability between different forms of giving". An LLC can make grants, investments and direct payments, while a charity faces stricter rules on what it can fund. The company also set a restriction: "None of the Anthropic equity held by the LLC will be used for lobbying or political giving, in the US or abroad." Governance details, including who will control the fund's decisions, will be published in the coming months, the company said.
Roughly a fifth of the company
Anthropic said the new commitment builds on earlier pledges. When the company was founded, its co-founders pledged to donate 80% of their equity. Anthropic also runs an equity donation match programme for employees, in which about half of staff take part, according to the company. Combined with the new fund, roughly 20% of Anthropic's shares will be dedicated to the public good, the company said, before any additional percentage added at valuation milestones. That estimate is the company's own and has not been independently audited.
An alternative to executive awards
Anthropic framed the escalating structure as a contrast with common pay practice in technology companies. "It is common practice for tech companies to offer escalating compensation awards to their executives," the statement said. "We are instead directing that equity to the public benefit." The comparison is pointed at a sector where large share grants tied to valuation targets have become controversial. Whether it changes how the market values Anthropic shares will depend on how much equity the fund eventually holds.
Why the timing matters
The fund arrives as Anthropic prepares for a stock market listing. Bloomberg reported, via Moneycontrol, that the company may start formally marketing its initial public offering as soon as the week of 9 November, citing people familiar with the matter. That would put trading on track to begin before the US Thanksgiving holiday. Shareholder approval for the 5% commitment would therefore come at a sensitive moment, when existing investors are assessing the value of their stakes ahead of a listing.
A second announcement in the same week
The fund was not Anthropic's only policy move of the week. On Friday, 16 October, Politico reported that Anthropic had quietly circulated its own draft AI safety bill to members of Congress. Anthropic confirmed the draft the next day, according to RTHK and the policy tracker AI Policy Landscape. The draft is dated 29 September and has no congressional sponsor. It is a proposal, not legislation under formal consideration.
What the draft bill would do
Under the draft, the federal government could block a model's release or use only if it poses a foreseeable, imminent risk of more than $1 billion in damages, or of killing or seriously injuring more than 50 people, Reuters reported, citing Politico. AI Policy Landscape said the Commerce Department would have to ask a federal judge in Washington for a 45-day blocking order and prove the risk through one of four pathways: biological weapons, loss of control, offensive cyber operations or automated AI research that amplifies those risks.
Pre-emption of state rules
The draft would also pre-empt state disclosure, third-party evaluation and incident-reporting obligations on AI developers, according to AI Policy Landscape. That would override parts of California's SB 53, the state's frontier AI transparency law. A legal analysis posted on LessWrong argued that, as written, the pre-emption clause does not stop states from suspending models. Anthropic's federal affairs lead, Anthony Cimino, said on 18 October that the company would revise the language so that states keep a role in evaluator standards and in regulating a lab's capacity to shut down a model.
| Provision | Draft text |
|---|---|
| Blocking order | Imminent risk of >$1bn damage or >50 deaths/serious injuries |
| Duration | 45 days, ordered by a federal judge in Washington |
| Incident report | 15 days, or 24 hours if imminent risk |
| Evaluator alert | 72 hours |
Obligations in the draft
The draft is not only about limits on government power. It would require every frontier developer to publish a transparency report at each new deployment and report critical safety incidents within 15 days, or within 24 hours if there is an imminent risk of death or serious injury, according to AI Policy Landscape. Large developers would have to publish safety frameworks, issue risk reports every 180 days and hire a registered independent evaluator. The evaluator would publish its findings and alert Commerce within 72 hours of finding conditions that could cause catastrophic harm.
Philanthropy and pre-emption side by side
The two announcements point in different directions for critics. The fund presents Anthropic as willing to share the gains from AI with people who may lose out. The draft bill would set a single federal standard that limits state oversight and allows a model to be stopped only at high harm thresholds. Supporters argue that a national standard prevents a patchwork of state rules. Critics argue that a company proposing its own regulator and thresholds is writing the rules it will be judged by. Both readings rest on the same facts.
The company's defence of the thresholds
Cimino defended the blocking-order standard as guarding a substantial power against abuse, according to AI Policy Landscape. He said the bill leaves other national-security authorities and export controls in place. He also said the thresholds were worth re-evaluating and that the company was developing stricter incident-reporting requirements and safeguards against "evaluator shopping", where developers choose lenient reviewers. Those statements indicate the draft is likely to change before any lawmaker adopts it.
Market context
The announcements came in a difficult week for AI shares. The broker XTB noted on Monday that signs of slowing revenue growth at Anthropic had added to investor nervousness during last week's selloff in AI-linked stocks. Anthropic is private, so its shares do not trade publicly, but sentiment about its growth affects the valuation of listed partners and of cloud companies that sell it computing power.
The open questions
Several details are still missing. Anthropic has not said who will govern the fund, how distributions will be decided or what the valuation baseline is. It has not said when stockholders will vote. On the draft bill, it has not published a revised text. The company said it will share more about the fund's structure in the coming months and report on its mission in an upcoming public benefit report. Until then, the size and use of the fund remain commitments rather than completed transfers.
Links to independent evaluation
Anthropic paired its week of announcements with a step on outside testing. AI News reported that the company is giving a team from METR, a nonprofit that evaluates frontier AI models, employee-like access, and that it is in early discussions with other organisations, including Hugging Face. Independent evaluation is also a central element of the draft bill. The company appears to be building a track record for the system of outside evaluators that its proposal would make mandatory for large developers.
What to watch
Three things will show how much weight to give the announcements. The first is the stockholder vote on the 5% commitment. The second is the governance structure, including whether outsiders sit on the fund's board. The third is whether any member of Congress sponsors a version of the draft bill, and whether its pre-emption clause survives. Each will show whether the week's moves change the rules for the industry or mainly shape perceptions ahead of the listing.
Journalistic Scoop
The LLC structure lets the fund do what a charity cannot, including paying people directly. The choice of a limited liability company over a charity is the detail with the most practical consequence. Anthropic said it wants to be able to explore "direct distributions to specific populations", something a US charity can do only under strict rules. An LLC owned by the company can, in principle, make payments to workers affected by automation, invest in retraining businesses or fund research, all from the same pool of equity.
The same flexibility means less outside oversight than a charity would face, which is why the governance details Anthropic has promised will matter. The one firm restriction published so far is that the fund's equity cannot be used for lobbying or political giving. That restriction separates the fund from the company's own policy work, including the draft federal bill it circulated in the same week.
Sources
- Anthropic, Investing in the public benefit, 18 Oct 2026 · anthropic.com
- CNBC, Anthropic to commit 5% stake to new public benefit fund, 18 Oct 2026 · cnbc.com
- Bloomberg via Moneycontrol, Anthropic to shift some shares into new public good fund, 19 Oct 2026 · moneycontrol.com
- Politico, Anthropic quietly pitches its own AI safety bill to lawmakers, 16 Oct 2026 · politico.com
- Reuters via Investing.com, Anthropic quietly pitches its own AI safety bill, 17 Oct 2026 · investing.com
- AI Policy Landscape, Anthropic Draft Frontier AI Bill · aipolicy.tech
Related
- OpenAI Discloses Four Cases of Its Models Breaking Through Test Boundaries, Including Forged Tokens and a Borrowed Wallet
- Yelp, Delta, JPMorgan and Schwab Shut the Door on Outside AI Agents as the Open Web Turns Into a Patchwork of Licences
- Eight Days Before Israel's General Election, Polls Show Netanyahu's Bloc Slipping and Neither Side Near a Majority