Most companies have a simple goal: make money for shareholders. From the start, Anthropic put a limit on itself. It made “developing advanced AI responsibly” a goal it must pursue by law, then let a group of outsiders who hold no shares gradually gain the power to pick directors.
This piece breaks down the Public Benefit Corporation (PBC), the Long-Term Benefit Trust (LTBT), and the reported voting setup after the IPO.
First things first: what is a Public Benefit Corporation (PBC)?
A Public Benefit Corporation (PBC) is a U.S. for-profit corporate form. Its charter requires the board to pursue shareholder interests while also weighing a defined public benefit. Anthropic’s is “responsibly developing and maintaining advanced AI for the long-term benefit of humanity.”
At an ordinary company, a director who gives up profit for a mission can be sued by shareholders. At a PBC, the charter authorizes directors to weigh the public interest. A PBC doesn’t make a company safe on its own. It only makes “looking beyond the share price” a legitimate basis for decisions. OpenAI also adopted this form for its for-profit entity after its 2025 restructuring.
The truly unusual part: the Long-Term Benefit Trust (LTBT)
Rarer still is the Long-Term Benefit Trust (LTBT), announced in 2023. Its members are outside trustees who hold no shares in the company. They answer to “the long-term interests of humanity,” not to the share price. Through a special class of stock (Class T), the trust gradually gains the power to appoint and replace directors.
The trust keeps governance power on the side of the safety mission. But shareholders with a large enough supermajority can still change its powers without the trustees’ consent.
As of October 2026, there are three trustees: Clinton Health Access Initiative CEO Neil “Buddy” Shah, Center for a New American Security CEO Richard Fontaine, and former Federal Reserve Chair Ben Bernanke, who joined in July 2026. Former trustee Mariano-Florentino Cuéllar became the company’s global affairs chief in August 2026, and the trust is looking for a replacement.
Is the trust already “live”?
Yes. In April 2026, Novartis CEO Vas Narasimhan joined the board through a trust appointment. Since then, directors appointed by the trust have been a majority of the board.
The trust shapes governance mainly by appointing and removing directors, and it gets notice of major changes. Strategy, products, and business decisions stay with management and the board.
After the IPO: the founding team’s voting power
Anthropic confidentially submitted a draft IPO filing (S-1) in June 2026. As of October 2026, the prospectus is not yet public. According to Reuters, which saw the prospectus in September, governance after the IPO splits like this:
| Role | Power |
|---|---|
| Long-Term Benefit Trust | Picks 4 of the 7 directors |
| Founder LLC (majority vote of co-founders) | Controls a single Class F share holding 50.1% of the votes on key company matters |
| Ordinary shareholders (Class A) | One vote per share; together with Class F, elects Dario, Daniela, and one other director |
The company stays a Delaware public benefit corporation. In other words, the trust keeps picking a majority of the directors, but the founding team holds a majority of the votes on key matters. The two now guard the company together.
The gatekeepers: the board and the founding team
The board is made up of Chair Daniela Amodei (President), CEO Dario Amodei, and four outside directors: Netflix co-founder Reed Hastings, former Microsoft CFO Chris Liddell, Vas Narasimhan, and Yasmin Razavi.
The founding team all came from OpenAI and left because they believed AI safety deserved to be taken more seriously. For each founder’s background and current role, see Dario Amodei and Anthropic’s founding team.
The biggest tension in this bet
On one side is fast commercialization: a private valuation of about $965 billion in May 2026, and, according to the prospectus Reuters saw, $518 billion in cloud, compute, and infrastructure commitments. On the other side, the company relies on the PBC, the trust, and its Responsible Scaling Policy to stay “safety-first.”
The two collide over concrete decisions. Outsiders usually watch three points:
- The line on defense use: in early 2026, Dario Amodei refused a demand from the U.S. Department of War (formerly the Department of Defense) to remove Claude’s safety protections. The Department of War then issued two “supply-chain risk” designations against Anthropic. A federal district court overturned one in August 2026. A federal appeals court upheld the other 2–1 in September, and Anthropic said it is considering further relief.
- Adjusting safety commitments: some media reported that Anthropic relaxed certain early safety commitments. The company responded with a new Responsible Scaling Policy that stresses more transparency and outside review.
- Restricted release: Claude Mythos 5.1, launched in September 2026, is the same model as Fable 5.1 with different safety protections, and is open only to vetted users. This “half-open” approach is itself a governance call.
Penchan’s summary
Anthropic is challenging a Silicon Valley default: that a company should only maximize shareholder value. It decided a founder’s promise isn’t reliable enough, so it wrote the mission into the charter and handed a non-shareholding trust the last line of defense.
What’s really worth watching is whether this structure works when interests and the mission collide head-on. How many safety slogans it has shouted matters less.
Further reading: What kind of company is Anthropic, Anthropic’s valuation and IPO, Anthropic’s risk list.