Silicon Valley’s artificial intelligence industry is increasingly experimenting with unconventional corporate governance models, raising new questions about accountability, investor rights, and public oversight.
Two of the world’s leading AI companies, Anthropic and OpenAI, have created corporate structures designed to protect their long-term missions. Instead of relying solely on traditional boards with standard fiduciary responsibilities, both companies have established independent “guardians” to oversee their broader objectives.
OpenAI, originally founded as a nonprofit organization, became a prominent example of the risks associated with complex AI governance in November 2023. Its board attempted to remove CEO Sam Altman, but the decision triggered a backlash from investors and employees. The dispute ultimately resulted in the replacement of most board members and a wider reorganization of the company.
Anthropic’s governance model is generally viewed as less volatile than OpenAI’s structure. Its long-term oversight mechanism includes a built-in “kill switch” that allows the trustee to be removed if holders of 85 percent of the company’s voting rights support the decision. That threshold could change if Anthropic eventually becomes a publicly traded company, according to people familiar with its corporate structure.
People close to Anthropic say investors who participated in multiple funding rounds understood the company’s governance arrangements before committing capital. Some investors reportedly viewed Anthropic’s focus on AI safety as a central part of their investment thesis.
However, one venture capitalist who invested in Anthropic said early backers also believed the company needed to grow into a major commercial business to achieve its long-term mission.
“Ultimately, investors decided that capitalism would prevail. If a company needs enormous amounts of money to develop and compete, investors will naturally assume that it must become a business,” the venture capitalist said.
As Anthropic faces growing pressure to generate profits, a potential initial public offering could expose its governance model to a broader—and potentially more demanding—group of shareholders.
Harvard University’s Fried wrote in a July paper that Anthropic’s current structure could give its directors greater freedom to prioritize the company’s mission over short-term financial performance. He noted that OpenAI has already experienced a significant governance failure, while Anthropic has avoided a comparable crisis because of its more stable structure.
Fried advised investors to carefully examine the governance agreements between the companies, which could change before an IPO, and to consider those arrangements when valuing any future public shares.
© 2026 The Financial Times Ltd. All rights reserved. Please do not copy and paste FT articles, redistribute them by email or post them on the web.
Source: arstechnica.com