Anthropic, the artificial intelligence company behind the Claude model, is preparing for an initial public offering (IPO) that could value the firm at up to $2 trillion, drawing attention to its unique governance structure centered around an external group of trustees. The Long-Term Benefit Trust (LTBT), a small advisory body established to ensure Anthropic’s mission prioritizes the long-term benefit of humanity, exercises significant influence over the company despite holding no equity.
The LTBT controls the majority of Anthropic’s board appointments, selecting four out of seven directors, including prominent figures such as Netflix co-founder Reed Hastings and Novartis CEO Vas Narasimhan. The trust currently has three members—Neil Buddy Shah, who chairs the group and leads the Clinton Health Access Initiative; former Federal Reserve Chair Ben Bernanke; and Richard Fontaine, CEO of the Center for a New American Security. Mariano-Florentino “Tino” Cuéllar, a former California Supreme Court justice, briefly served on the trust before becoming Anthropic’s chief global affairs officer.
Anthropic has positioned the LTBT as a potential governance model for future AI companies, aiming to establish industry standards that balance commercial objectives with ethical considerations, akin to accounting standards like GAAP. Trustees receive advance notice of major corporate actions, including new AI model launches, and engage regularly with company leadership. This oversight has included discussions around the controlled deployment of the Mythos cybersecurity model and negotiations with the U.S. government over automated weapons technology.
Despite the trustees’ extensive involvement, they have primarily played an advisory role and have not yet faced scenarios requiring them to enforce strict limits on the company’s pursuit of profit versus its mission. This has led some experts to voice concerns about inherent conflicts in the governance structure. Harvard Law School professor Jesse Fried highlighted tensions in allowing profit-seeking investors to fund the company while placing mission control in the hands of unelected trustees who may prioritize goals other than profitability. He noted that such governance arrangements embed substantial tensions that have yet to be fully tested.
Other experts acknowledge the thoughtful design of Anthropic’s governance but caution that its effectiveness remains unproven, particularly under increasing commercial pressures and amid intense competition within the AI sector. Elizabeth Pollman, a law professor at the University of Pennsylvania, emphasized the difficulty of creating governance frameworks capable of managing competing interests as a company grows and operates in a complex global environment.
Anthropic’s model continues a broader trend among leading AI firms such as OpenAI, which have departed from traditional corporate governance norms by appointing “guardians” focused on mission oversight. However, Anthropic’s trust differs from OpenAI’s board structure by incorporating a “kill switch,” allowing trustees to be removed by an 85 percent shareholder vote—a threshold that may adjust upon the company’s public listing.
Private investors backing Anthropic have reportedly accepted the governance framework, recognizing the company’s emphasis on safety as integral to their investment rationale. At the same time, some investors acknowledge the necessity of establishing a commercially viable business to support Anthropic’s ambitions, reflecting an underlying belief that market success is essential to sustaining its mission.
With the transition to public ownership, Anthropic’s governance model will face increased scrutiny from a wider investor base focused on its financial performance. Experts caution that investors should carefully evaluate the company’s governance arrangements, which may evolve ahead of the IPO, considering the unresolved challenges inherent in balancing profit motives with long-term societal goals in a rapidly advancing and competitive AI landscape.