How Anthropic Plans to Shield Its Safety Mission From Wall Street After an IPO

Anthropic wants to tap the public markets through an initial public offering while insisting it will not give up its core mission of developing AI safely and for the benefit of humanity. That goal carries a tension: more capable models bring greater risks, and shareholders focused on returns may have little patience for costly safety work that slows products on their way to market.

OpenAI CEO Sam Altman has cited investor pressure as one reason his company is not pursuing an IPO this year. Anthropic is betting that an unusual governance design can keep such pressure in check. University of Pennsylvania law professor Elizabeth Pollman told The Verge that the company has proposed going public with a "unique and complicated governance structure that has not been used before" and that was "creatively designed for this founder-led and mission-driven company."

The first layer is legal status. Anthropic is a public benefit corporation, which means it must weigh more than shareholder returns, balancing profit against its stated social mission and the interests of stakeholders such as employees and specific communities. Patagonia and Kickstarter are among the better-known examples of the model.

The second layer limits shareholder influence over the board of directors, which oversees the company and appoints its leadership. Shareholders normally hold considerable sway over who sits on a board, and could replace directors if they objected to what they saw as needless delays from optional safety testing. Under Anthropic's plan, four of the seven directors will be chosen by a body called the long-term benefit trust, according to Jens Dammann, a business law professor at the University of Texas School of Law.

The trust consists of five financially disinterested voting trustees with expertise in AI safety, national security, public policy and social enterprise. Through a special class of shares known as Class T stock, they can appoint or remove directors to ensure the company responsibly balances stockholder interests with those affected by its conduct and its public benefit purpose.

A third safeguard covers the remaining three seats. Ordinary shareholders technically vote on them, but an entity called Founder LLC will hold 50.1 percent of the voting power on shareholder matters. Because many votes are decided by simple majority, the founders can effectively block or force through decisions, and Columbia Law School professor Dorothy Lund said they will in practice choose those directors. The prospectus says seven founders will initially be the sole members: Dario Amodei, Daniela Amodei, Tom Brown, Jack Clark, Jared Kaplan, Sam McCandlish and Christopher Olah. Cofounder Ben Mann, who co-leads the experimental Anthropic Labs arm, is not among them, The San Francisco Standard reported.

Dammann said the arrangement gives the founders and the trust "tremendous power" and shields the company from many pressures of being public. Ann Lipton, a University of Colorado Law School professor, agreed that legally the company is "entirely insulated from profit seeking pressures," though she noted the founders have their own financial stakes that could influence their choices.

Legal insulation is not the same as immunity from the market, the experts cautioned. Lipton said AI companies "desperately need capital" and that no governance structure guarantees Anthropic can fund its operations unless it pursues profit opportunities faster than competitors. Lund said the company must still stay in its investors' good graces, adding that competitive pressure and capital needs can erode safety commitments over time. Anthropic did not respond to The Verge's request for comment.

Founder departures are another potential weak point. The prospectus calls the company's success highly dependent on its senior team, singling out Dario and Daniela Amodei as critical. Founder LLC would "sunset" if a majority of its members leave or the group dissolves it. Dammann said ordinary shareholders could then gain the power to weaken the trust's protections, even over the trustees' objections, and other routes exist, including coordinated action by public shareholders, employee stockholders and founders.

For now, Anthropic can probably pursue both goals, but the protections are far from ironclad. Dammann described a "fundamental dilemma that cannot be resolved in a fully satisfactory way." He added: "Investors may not mind their lack of voice as long as things go well. But if things go sour, they may become quite frustrated."