Anthropologie is not a traditional commercial company. Since its inception on January 26, 2021, this Artificial Intelligence (AI) laboratory Registered as a PBC (Public Benefit Corporation or Public Benefit Corporation). This declaration of intent means that its directors have an obligation to balance the economic interests of shareholders with those affected by the company’s activities. Furthermore, its statutes state that its purpose is to contribute to the public good.
Dario Amodei, CEO of Anthropic, and other spokespeople for this company have declared many times that their reason for existence is to develop and maintain AI systems in a responsible way and guarantee the benefit of humanity in the long run. That all sounds great, but the lab’s IPO, scheduled for mid-November, could undermine the principles its founders clung to when they left OpenAI and founded it.
Anthropic plans to retain leadership decision-making power when the company goes public. Shareholder pressure could threaten its founding principles, and to avoid this the seven co-founders, including siblings Daniela and Dario Amodei, created a structure that would allow them to jointly control 50.1% of voting rights in certain corporate decisions. The device they are going to use to accomplish their goals is very clever.
Following in Palanti’s footsteps
Anthropic is going to separate administrative and strategic control of the company from economic ownership of its shares. The process that will make this plan possible revolves around Founder LLC (Founder Limited Liability Company or Founder’s Limited Liability Company), a limited liability company that will bring together Anthropic CEO Dario Amodei, his sister Daniela Amodei, the company’s president, and five other co-founders.
This design fits like a glove with the founders’ pledge to donate 80% of their financial participation to charity.
This entity will receive a single class F share, which is nothing more than a special share that will consolidate 50.1% of the voting power in decisions submitted to shareholders. This process is similar to one used by Palantir to protect its founders’ control, so it’s not really a new strategy. This is relatively uncommon though. Dario Amodei will have a relatively modest economic stake, but the voting structure will allow him to exercise far more influence than is commensurate with his ownership percentage.
On the other hand, this design fits like a glove with the founders’ pledge to donate 80% of their financial participation to charity. In addition, Anthropic is considering the possibility of offering its employees a special type of shares that would help them break tied voting ties. This decision, if ultimately carried out, will support the balance of its governance structure, although founder control will remain a central element.
Either way, Anthropic’s plan involves risk. A firm hold on control of the laboratory that the founders plan to maintain may deter some potential shareholders, who may not be interested in acquiring one. Very limited impact power. On the other hand, going public will not eliminate the tension between the need to be profitable and the need to keep AI safe. Its framework seeks to allow senior management to forego commercial opportunities to prioritize safety, but it is unclear how shareholders will accept that such decisions harm their short-term financial results.
Picture | TechCrunch (edited)
More information TechCrunch
In Xataka | “I don’t trust Anthropic and OpenAI to open frontier AI”: A Dipsic engineer complains to the US
