The Washington Post flags Anthropic's weak spot: asking investors to trust seven founders to govern AI

By The Washington Post · October 28, 2026.

The Washington Post editorial board published an opinion column on October 28 about the governance structure Anthropic will have when it goes public. Its thesis: ordinary shareholders will have "unusually little say" in the company's governance.

According to the piece, Anthropic argues that artificial intelligence is too consequential a technology to be governed solely by ordinary commercial incentives. To protect the public interest, the company says its seven founders are "uniquely qualified to be stewards of our mission." The Post counters that the company undervalues the virtues of competition and market discipline.

A note on the source: the material we had access to is only the opening of the column, with the headline, the standfirst and the first few lines of text. We do not know its full arguments or the specific figures of the control arrangement being criticized. We therefore limit ourselves to what the text says and to our own reading.

**Our take.** Both sides touch on a real problem, and neither fully solves it.

The criticism carries weight. Asking for public capital in exchange for little control is a familiar formula in the tech world, and "trust us" is a weak guarantee when the stakes are described as civilizational. Market discipline is not perfect, but it forces accountability. Founders, however well-intentioned, will not necessarily remain so over time or under mounting financial pressure.

Even so, Anthropic's argument is not trivial. In general, quarterly market pressure can push companies to cut corners on safety, and enhanced-control structures exist precisely to shield a long-term mission. The problem is that the shield protects those who use it well and those who don't alike. Those who defend it are betting on people; those who criticize it, on checks and balances.

As industry context, this tension is sharpened because frontier labs are capital-intensive businesses, with compute commitments worth hundreds of billions of dollars. The more they depend on outside financing, the harder it is to argue that investors should accept concentrated control without a voice.

Our position: the answer is not to choose between founders and the market, but to demand verifiable governance. That includes boards with independent members and real power, transparency about how safety decisions are made, and accountability mechanisms that do not depend on anyone's good faith. It is the same pattern we advocate for AI agents: trust based on auditable records, not on statements of intent.

In the long run, the horizon of abundance and eradicated diseases requires that the technology be governed legitimately. If Anthropic's IPO (initial public offering) sets a precedent, it matters that it be one of a mission protected by checks and balances, not of a promise that can only be believed or not.