Anthropic opts for Frog and Toad governance

Space Exploration Technologies

30/10/2026 16:37

The author is a Reuters Breakingviews columnist.  The opinions expressed are his / her own.
By Robert Cyran
NEW YORK, Oct 30 (Reuters Breakingviews) - Anthropic and its ilk predict that artificial intelligence will soon hand unfathomable power to humanity. That naturally raises fears of model makers misusing superintelligence or allowing the genie out of the bottle. For SpaceX SPCX.O boss Elon Musk or Meta Platforms META.O Chief Executive Mark Zuckerberg, the solution is dictatorial control: they steer their companies through super-voting shares, meaning they can guide events unequivocally through what one hopes is sober reflection. Anthropic’s founders similarly don’t want to share influence with outside investors after its planned listing, yet they have put in some questionably durable safeguards. Call it Frog and Toad governance.
In the story “Cookies” by children’s author Arnold Lobel, the two amphibians face a dilemma. They have some delicious treats, but want to stop themselves from eating too many and getting sick. Frog puts them in a box; Toad points out that they can open the box. He keeps adding safeguards, tying it up and putting it on a high shelf. Of course, he can just get it down and open it up again.
Anthropic has set up a more complicated version. It is chartered as a public benefit company and required to balance the interests of shareholders and society. Seven co-founders control a single special share, giving them 50.1% of the total voting power at the Claude maker. They can also appoint a minority of members of the company’s board. As a counterweight, a “Long Term Benefit Trust” composed of five trustees can appoint or remove a majority. The idea is members like Ben Bernanke will use the Trust to balance the company’s goals against short-term drives.
Yet when the Trust nominates or attempts to remove one specific director, a three-person committee must approve the decision. The founders control two seats on this committee. If the two sides can’t agree, then the decision is thrown to Class E shareholders, an otherwise largely powerless form of stock given to employees.
In sum: the founders will steer this ship. The rules, Anthropic’s status as a public benefit company, and sunset provisions that end the founders’ power under certain conditions serve mostly as nudges for executive behavior. After all, even without special classes of stock, CEOs and founders usually have outsized say. Steve Jobs, for example, had absolute say over Apple's AAPL.O direction, despite owning only a small chunk of stock.
Frog realized the only way to avoid temptation was feeding cookies to the birds. CEO and co-founder Dario Amodei has repeatedly said opting out of the AI race isn’t an option. That leaves nothing but willpower, bucked up by the friction of governance inconveniences.
Follow Robert Cyran on Bluesky.
     
CONTEXT NEWS
Artificial intelligence lab Anthropic ​filed the registration statement for its planned initial public offering on October 26.

(Editing by Jonathan Guilford; Production by Pranav Kiran)
((For previous columns by the author, Reuters customers can click on CYRAN/robert.cyran@thomsonreuters.com; Reuters Messaging: robert.cyran.thomsonreuters.com@reuters.net))