Harvard Research on Anthropic Governance Risks

Harvard Law School professor Jesse Fried and Idan Reitter published a paper analyzing Anthropic’s governance structure ahead of a potential IPO. The authors compare Anthropic’s model to that of Ben and Jerry’s and OpenAI, highlighting risks associated with independent bodies that can limit investor decisions.

Unlike OpenAI, where a nonprofit foundation maintains full control, Anthropic’s structure includes a protective mechanism for investors. A supermajority of Anthropic shareholders can dissolve the Long Term Benefit Trust and remove appointed directors. Researchers describe this mechanism as a safeguard that makes the system more resilient to conflicts.

The authors note that this model remains untested under real market pressure.