The Claude developer gives a long-term benefit trust a role in electing its board alongside stockholders, a design worth watching as Canada sets its own course on AI.
Anthropic governance is built to keep the company’s safety mission from being traded away for short-term returns. The San Francisco–based developer of the Claude AI models has set out how its board is chosen, who sits on it and which values guide its decisions. For Canadian businesses adopting AI tools and for policymakers weighing how to regulate them, the structure is one of the clearest examples of a frontier AI lab writing its priorities into its corporate design.
Anthropic describes itself as a public benefit corporation. Its stated purpose is the responsible development and maintenance of advanced AI for the long-term benefit of humanity. Its directors are elected by two groups: ordinary stockholders and a separate body called the Long-Term Benefit Trust.
The company’s own materials say little about operations, staffing or partnerships in Canada. This article reports what Anthropic has disclosed, then sets out clearly labelled context on why the model matters to Canadian readers.
How Anthropic Governance Splits Board Elections Between Stockholders and a Benefit Trust
Most companies answer to their shareholders, and directors are judged on financial performance. A public benefit corporation works differently. It names a public purpose in its charter, and its leadership is expected to weigh that purpose alongside investor interests.
Anthropic goes a step further. Stockholders do not elect the whole board on their own. The Long-Term Benefit Trust, often shortened to the LTBT, also takes part in electing directors. The arrangement is meant to keep people focused on the company’s long-term mission inside the boardroom, even as investment and commercial pressure grow.
The company’s public summary does not say how many seats each group fills, how trustees are appointed, or how any disputes between the trust and investors would be settled. Those details are not specified in the material reviewed for this report.
Six Directors and Four LTBT Trustees: Who Holds the Seats
Anthropic lists six people on its Board of Directors and four trustees on the Long-Term Benefit Trust. The company does not give their individual roles or committee duties.
| Body | Members listed by Anthropic |
|---|---|
| Board of Directors | Dario Amodei, Daniela Amodei, Yasmin Razavi, Reed Hastings, Chris Liddell, Vas Narasimhan |
| Long-Term Benefit Trust trustees | Dr. Ben Bernanke, Richard Fontaine, Neil Buddy Shah, Dr. Christopher Stubbs |
General context: Several names will be familiar to business readers. Reed Hastings is best known as the co-founder of Netflix. Ben Bernanke is a former chair of the U.S. Federal Reserve. Dario Amodei and Daniela Amodei are Anthropic co-founders.
The trustee list is weighted toward economics, public policy, global health and science rather than venture investing. That mix reflects the trust’s role as a counterweight focused on the mission. Boardroom composition is drawing more attention across sectors. Recent appointments such as Enoch Kariuki joining the Epicrispr Biotechnologies board show how companies in fast-moving, high-risk fields are shaping their oversight bodies.
Seven Values Anthropic Says Should Settle Hard Calls
Beyond the formal structure, Anthropic sets out seven values it says guide everyday decisions. In its own framing, the mission is the final arbiter. Paraphrased, the values are:
- Act for the global good: make decisions aimed at the best long-run outcomes for humanity, and be willing to act boldly so the technology remains a positive force.
- Hold light and shade: accept that AI could bring unprecedented harm or unprecedented benefit, and plan for both.
- Be good to users: a broad definition of “users” that covers customers, policymakers, employees (whom the company calls “Ants”) and anyone affected by its technology.
- Ignite a race to the top on safety: push competitors to match or beat its safety and security standards.
- Do the simple thing that works: favour empirical, iterative fixes over complex ones. The company puts it as not building a spaceship when a bicycle will do.
- Be helpful, honest and harmless: run a high-trust, low-ego culture that communicates directly and repairs relationships when needed.
- Put the mission first: treat the mission as the shared purpose that resolves competing goals, with each staff member taking personal ownership of it.
Values statements are common in corporate life. What stands out here is that the governance design is meant to reinforce them. The trust gives the mission a formal voice in who sits on the board.
The “Race to the Top on Safety” Pitch and What It Asks of Rivals
The most outward-facing of the seven values is the call for a “race to the top” on safety. Anthropic argues that reliable, trustworthy and secure AI systems are a shared responsibility of the industry. It also argues that the market will reward companies that build them.
The aim is to create competitive pressure in which developers try to outdo one another on safety and security rather than only on capability or speed. Anthropic says it wants to keep setting that bar and to push others to follow.
The company frames AI safety as a systematic science. It describes a loop in which research findings are applied to products, lessons from deployment feed back into research, and results are shared publicly. Its stated goal is frontier AI that is reliable, interpretable and steerable, meaning systems whose behaviour can be understood and directed.
Security is part of the same promise. For Canadian organisations, the question is not abstract: Canada’s Cyber Centre recently urged urgent patching against PHP2Shell attacks. That alert is a reminder that the software sitting beneath AI tools is a target in its own right.
Research, Policy, Product and Operations: The Four Teams Behind Claude
Anthropic organises its work into four broad functions:
- Research: frontier AI work that ranges from interpretability and alignment to AI’s economic and social effects.
- Policy: sharing what the company observes at the frontier with policymakers and civil society in the United States and abroad.
- Product: turning research into practical tools such as Claude for businesses, non-profits, civil society groups and individuals worldwide.
- Operations: people, finance, legal and recruiting teams. The company notes that staff have come from NASA, start-ups and the armed forces.
The company also says it sees itself as one part of a wider effort. It describes working with civil society, government, academia, non-profits and industry to promote safety across the field. Anthropic invites people to apply through its open roles listing. The company materials reviewed do not specify any Canadian locations, Canadian hiring numbers or Canada-based teams.
Why a Benefit-Trust Model Matters for Canada’s AI Debate
Context, not drawn from Anthropic’s disclosures: Canada has a significant place in AI research history. The country is home to well-known institutes, including Mila in Montreal, the Vector Institute in Toronto and Amii in Edmonton. Many Canadian companies now use AI models built by U.S. labs, which makes those labs’ internal checks relevant to Canadian customers.
Canadian AI regulation is still unsettled. The proposed federal Artificial Intelligence and Data Act, part of Bill C-27, did not become law before Parliament was prorogued in early 2025. Readers should check the current status of any successor legislation, as federal policy in this area continues to change.
Until clearer rules arrive, voluntary corporate structures carry extra weight. Anthropic’s model offers Canadian lawmakers and boards a concrete case study. It raises three questions:
- Can a trust with a role in board elections hold a company to its stated mission when commercial pressure rises?
- Should procurement teams in Canada ask AI vendors about their governance, not just their pricing and performance?
- Could public benefit or trust-style oversight become a model for Canadian AI start-ups seeking investors?
For business leaders buying AI services, a practical step is to treat a vendor’s governance the same way they would treat its security certifications. Ask who oversees it, how its board is chosen, and what commitments are written into its charter.
Frequently Asked Questions
What is a public benefit corporation?
It is a company that commits to a stated public purpose alongside earning returns for investors. Anthropic’s stated purpose is the responsible development and maintenance of advanced AI for humanity’s long-term benefit.
What does Anthropic’s Long-Term Benefit Trust do?
Anthropic says its board is elected by stockholders together with the Long-Term Benefit Trust. The trust’s four listed trustees are Ben Bernanke, Richard Fontaine, Neil Buddy Shah and Christopher Stubbs.
Who sits on Anthropic’s board?
The company lists Dario Amodei, Daniela Amodei, Yasmin Razavi, Reed Hastings, Chris Liddell and Vas Narasimhan.
Does Anthropic have offices or staff in Canada?
Not specified. The company materials reviewed for this article do not mention Canadian operations.
