Anthropic has announced the creation of the Long‑Term Benefit Trust (LTBT), a new governance structure the company has been developing since its founding. The LTBT is intended to fine‑tune Anthropic’s corporate governance to address the long‑term risks and societal impacts that transformative artificial intelligence (AI) may produce.
What the LTBT is and how it works
The LTBT is an independent body of five Trustees who are financially disinterested and bring expertise in AI safety, national security, public policy, and social enterprise. At the close of Anthropic’s Series C round, the company amended its corporate charter to create a new class of stock, Class T, held exclusively by the Trust. Class T stock gives the Trust the authority to elect and remove an increasing number of Anthropic board members according to time‑ and funding‑based milestones; the Trust will elect a majority of the board within four years.
Class T also includes protective provisions that require the Trust to receive notice of certain actions that could materially alter the company or its business. The Trust is organized as a purpose trust under Delaware common law with the same public‑benefit purpose stated in Anthropic’s charter: the responsible development and maintenance of advanced AI for the long‑term benefit of humanity. The Trust must use its powers to ensure Anthropic responsibly balances stockholder financial interests with the interests of those affected by the company’s conduct and the public benefit purpose.
Why Anthropic says this is needed
Anthropic argues that standard corporate governance mechanisms tend to prioritize stockholders’ financial interests, while AI may create unusually large externalities—ranging from national security risks and large‑scale economic disruption to existential threats and major public goods in health and safety. Because legal rules and social norms that constrain other high‑externality activities have not fully caught up with AI, Anthropic decided to adjust its governance to better manage these potential externalities.
The company also notes that for most day‑to‑day decisions public benefit and commercial success are not in conflict and are often mutually reinforcing: safety research benefits from building frontier models, which in turn depend on commercial resources. Anthropic does not expect the LTBT to intervene in ordinary commercial strategy; rather, the Trust is designed to focus on high‑stakes, long‑range questions where the consequences extend well beyond the firm.
Relation to Anthropic’s Public Benefit Corporation status
Anthropic is incorporated in Delaware as a Public Benefit Corporation (PBC), a form that permits directors to balance stockholder value with a specified public benefit and the best interests of those materially affected by the corporation. While the PBC form gives the board legal latitude to weigh long‑ and short‑term externalities, it does not by itself make directors directly accountable to other stakeholders or align their incentives with the broader public. The LTBT adds a structural mechanism to supply accountability and incentives aimed at balancing stockholder interests and the public benefit at critical junctures.
An experimental governance instrument
Anthropic describes the LTBT as an experiment: its design was developed with input from prominent corporate governance scholars and lawyers and “red‑teamed” before implementation. The Trust’s powers phase in over time to allow course correction. Amendment procedures aim to balance resilience against attempts to bypass the structure with the need for flexibility while the Trust’s experimental nature is evaluated. There are failsafe provisions that permit changes to the Trust without Trustees’ consent if sufficiently large supermajorities of stockholders agree; the required supermajorities grow as the Trust’s power phases in.
Initial Trustees and subsequent changes
The initial Trustees named were:
- Jason Matheny — CEO, RAND Corporation
- Kanika Bahl — CEO & President, Evidence Action
- Neil Buddy Shah — CEO, Clinton Health Access Initiative (Chair)
- Paul Christiano — Founder, Alignment Research Center
- Zach Robinson — Interim CEO, Effective Ventures US
Trustees serve one‑year terms and future Trustees are to be elected by the Trustees. Anthropic conducted a year‑long search for this founding group. The company’s updates note several changes to the Trust’s composition after the original announcement: Jason Matheny stepped down in December 2023 to avoid potential conflicts with RAND Corporation policy work; Paul Christiano stepped down in April 2024 to take a role as Head of AI Safety at the U.S. AI Safety Institute; Kanika Bahl and Zach Robinson stepped down in January 2026 to pursue nonprofit and philanthropic work. New members joined: Richard Fontaine, CEO of the Center for a New American Security, in May 2025; Mariano‑Florentino (Tino) Cuéllar, President of the Carnegie Endowment for International Peace, in January 2026; and Dr. Ben Bernanke, Distinguished Fellow at the Brookings Institution and former Chair of the Federal Reserve, in July 2026.
Conclusion
Anthropic’s Long‑Term Benefit Trust represents a novel corporate governance experiment meant to provide an independent channel for public‑interest considerations in board composition and high‑stakes decision‑making as AI systems approach transformative capabilities. Through a new class of stock, protective provisions, phased powers and amendment safeguards, Anthropic aims to test whether this institutional design improves decision‑making when the societal stakes are highest.
