Anthropic wants to raise money from public markets without letting shareholders dictate how fast it ships or how much safety testing it does. A new analysis from The Verge, published October 27, 2026, asks whether that is possible. Its answer, from the legal scholars it quotes, is roughly: legally yes, practically only up to a point.
This post explains the structure in plain terms, shows where the experts disagree on the risks, and flags one detail where the Verge's description and the prospectus wording differ. It builds on our earlier coverage of the S-1 filing and the Long-Term Benefit Trust.
Quick reference: the four safeguards
| Safeguard | What it does | Source |
|---|---|---|
| Public benefit corporation | Legally required to balance profit with a social mission and stakeholder interests | The Verge |
| Class T stock and the Long-Term Benefit Trust | Trust appoints or removes four of seven directors | The Verge, citing the prospectus |
| Founder LLC with Class F stock | Controls 50.1% of the vote on shareholder matters | The Verge; the filing, as summarized in our S-1 explainer |
| Employee Class E shares | Non-economic shares, per our S-1 read | Our S-1 explainer |
Why the question is being asked
The Verge frames the tension plainly: greater model capabilities bring greater risks, and shareholders eager for returns may have little patience for costly safety measures that slow products on the way to market. It notes that OpenAI CEO Sam Altman has cited investor pressure as one reason OpenAI is not pursuing an IPO this year. We looked at Altman's remarks and others in our piece on three endgames for AI safety.
The concern is not hypothetical for Anthropic. New York City's comptroller has already written asking it to pause the IPO over safety governance, which we covered in the Levine letter post.
How the structure works
Public benefit corporation. Anthropic is a PBC, meaning it is legally required to focus on more than maximizing shareholder returns. The Verge points to Patagonia and Kickstarter as better-known examples. Directors can defend decisions that weigh safety against profit.
The Long-Term Benefit Trust and Class T. The prospectus, as the Verge describes it, gives the trust five "financially disinterested voting trustees" with expertise in "AI safety, national security, public policy, and social enterprise." Through Class T stock they can appoint or remove directors, with the stated purpose of ensuring Anthropic "responsibly balances the financial interests of stockholders with the interests of those affected by Anthropic's conduct and our public benefit purpose." University of Texas law professor Jens Dammann says this means four of the seven board seats are chosen by the trust.
Founder LLC. The remaining three seats are technically open to ordinary shareholders, but a Founder LLC will control 50.1% of voting power on shareholder matters. Because many votes are decided by simple majority, Columbia Law School professor Dorothy Lund told The Verge the founders will effectively choose those three directors. The prospectus says the seven co-founders will "initially" be the only members.
Put together: ordinary public shareholders have almost no say over the board. Colorado law professor Ann Lipton put it this way: "Anthropic is entirely insulated from profit seeking pressures, in the sense that ordinary shareholders have no ability to select the board or make business decisions."
What could still go wrong
The experts are not saying the structure fails. They are saying legal insulation and market reality are different things.
- Capital needs. Lipton said Anthropic "desperately needs capital" and that no governance structure protects against the reality that unless it takes profit-making opportunities at a faster pace than competitors, it cannot fund its operations. Lund said competitive pressure and capital needs can erode safety commitments over time.
- Founder financial stakes. Lipton also cautioned that the founders have their own financial interests in the company, which could influence their decisions even though shareholders cannot force them.
- Founder departures. The prospectus says success is highly dependent on senior leadership, especially Dario and Daniela Amodei. The Verge reports that if the Founder LLC sunsets, ordinary shareholders could gain the power to weaken the trust's protections, even against the wishes of trustees, and without the support of founders or employees, who hold different stock classes. Dammann says other routes exist, such as coordinated action by public shareholders, employee stockholders and founders.
- Investor patience. Dammann: "Investors may not mind their lack of voice as long as things go well. But if things go sour, they may become quite frustrated."
Dammann also sees a "fundamental dilemma that cannot be resolved in a fully satisfactory way" between the mission and investor demands, while saying he does not see how going public by itself would undermine Anthropic's ability to prioritize safety. Anthropic did not respond to The Verge's request for comment on whether it can resist pressure to accelerate at the cost of safety.
A discrepancy worth knowing: when does the founder vote end?
The Verge says the Founder LLC would "sunset" if a majority of its members leave or the group chooses to dissolve it. Our own read of the filing, in the S-1 explainer, says the Class F share ends when there have been fewer than three Continuing Founders for 90 consecutive days, followed by a 450-day transition period. These may describe different triggers, or the Verge may be summarizing loosely. If you are making a decision that depends on it, read the relevant section of the S-1 itself rather than relying on either summary.
Reading the terms: a short glossary
- Class T share. A special share held by the Long-Term Benefit Trust. One share is enough to elect four of seven directors, per our S-1 read, which is why the trust matters far more than its zero economic stake suggests.
- Class F share. The founders' voting share, worth 50.1% of the vote on shareholder matters. It is a control device, not an ownership claim.
- Controlled company. Our S-1 explainer notes that Anthropic would be a Nasdaq controlled company, a status that exempts a firm from some board-independence listing rules because one holder controls most votes.
- Public benefit corporation. A for-profit company with a legally stated public purpose. The legal effect is mostly to protect directors who weigh the mission against profit, and to let certain stakeholders raise concerns; it does not by itself guarantee any particular safety practice.
- Sunset. A clause that ends a special right under stated conditions. Both the exact trigger and the length of any transition period matter, which is why the discrepancy above is worth checking in the filing.
Taken together these explain the experts' split view. Each device removes one route by which shareholders could force a change: they cannot pick most directors, cannot outvote the founders, and cannot easily sue over a mission-driven decision. What none of them removes is the company's need to keep raising money in a market where competitors are moving fast, which is why Lipton and Lund focus on capital rather than on voting rights.
How this compares with other labs
Anthropic's setup is one of several experiments in tying a frontier lab to a mission. Our comparison of how AI labs structure public-benefit giving shows that LLCs, foundations and PBCs do different jobs. The distinctive part of Anthropic's design is not the PBC, which several firms use, but the combination of trust-picked directors and a founder supermajority, which the Verge calls highly unusual for a major technology company. Legal scholar Elizabeth Pollman, quoted in the article, called it a "unique and complicated governance structure that has not been used before."
What this means for builders and buyers
- Customers. If you rely on Claude, the governance question is really a continuity question: will the lab keep prioritizing safety processes, such as testing before release, even when competitors move faster? Release cadence is the observable proxy.
- Developers. Safety-related policy changes, usage restrictions and model-access decisions are board-level decisions at a company like this. Watch for them in the prospectus's risk factors and in later filings.
- Policy watchers. Whether the structure holds will be a template question for other labs. If Anthropic lists, others may copy parts of it, or avoid listing as OpenAI says it is doing this year.
- Investors reading the numbers. The business side, including reported revenue and losses, matters because capital need is exactly what the experts say could pull against the mission.
Questions to ask as the IPO nears
- Are the trust's governing documents ever fully disclosed? Reporting we covered earlier said they were largely not.
- Do any proposals to amend the stock classes surface after listing?
- How does Anthropic describe, in its own risk factors, the circumstances in which the trust's powers could be weakened?
- Does release cadence change, and does the company explain safety-related delays?
Related reading
- Anthropic S-1 explained
- Anthropic's Long-Term Benefit Trust: WSJ report vs public record
- NYC Comptroller asks Anthropic to pause its IPO
- How AI labs structure public-benefit giving
- Anthropic draft prospectus numbers
- Sam Altman on three endgames for AI safety
Details are accurate as of October 27, 2026. This is an explainer, not legal or investment advice.