Anthropic acknowledges the risks AI poses to humanity, but is its governance really up to the task?
Anthropic recognises the risks to humanity – but is its governance really up to the job? | Nils Pratley
In a Guardian column published before Anthropic's $1.4 trillion listing this week, Nils Pratley questioned whether its governance structure can check power. The article explains that Anthropic has two centers of power: the seven founders collectively hold one vote, accounting for 50.1% of the final voting power; the Long-Term Benefit Trust nominates four of the seven directors, and its trustees include former Federal Reserve Chair Ben Bernanke; the company is registered as a public benefit corporation. The author argues that the motive is not merely the founders seizing power; given that the listing documents state AI development could pose catastrophic or existential risks to humanity, unconventional governance has some justification, but it remains an untested experiment that may just be a more refined "trust us" structure. He quotes Stuart Riddick of Aberdeen: agreeing with the mission in principle is not the same as agreeing with the board's interpretation and execution. The author questions whether the trustees would publicly disclose a conflict with the founders over basic safety issues, and argues that the arrangement still represents a high concentration of power in a few individuals.