Harvard Law: Anthropic is about to sell a safety mission Wall Street can veto (fortune.com)

A recent Harvard Law paper by Professor Jesse Fried and S.J.D. candidate Idan Reiter analyzes the complex governance structures of AI firms, particularly in light of Anthropic's confidential IPO filing. The study underscores the risks associated with so-called "mission guardians," like those at OpenAI and Ben & Jerry’s, noting that these structures often fail to protect investors and can even undermine the intended missions. The authors highlight the Ben & Jerry's precedent, where the efforts of mission guardians to maintain social objectives ultimately harmed both the brand and its parent company, Unilever, resulting in a significant market value loss and internal conflicts. As Anthropic prepares for its IPO, it introduces a governance model that includes a "kill switch," allowing a supermajority of investors to terminate its mission trust and remove its appointed directors, providing a crucial corrective mechanism that was absent in OpenAI’s structure. While the paper suggests that this arrangement may offer a more viable path forward, it remains untested, raising questions about its effectiveness under pressure. The ongoing scrutiny of OpenAI's governance, particularly following its tumultuous recent history with leadership and safety concerns, highlights the critical need for sustainable governance models in the fast-evolving AI landscape.