Nobel laureates urged not to believe in Anthropic’s promises – but investors did not hear them

Источник изображения: ИИ-генерация ChatGPT/3DNews

The day before Anthropic’s triumphant listing on the securities exchange, a group of Nobel laureates and a number of heads of tech giants published an open petition accusing Anthropic of abandoning previously stated principles of safety in the development of artificial intelligence. According to the activists, this was supposed to disrupt the company’s debut, but the securities market remained deaf to the call – Anthropic shares were sold out like hot cakes in the winter.

  Image source: AI generation ChatGPT/3DNews

Image source: AI generation ChatGPT/3DNews

The statement, entitled “MisAnthropic”, which can be translated into Russian as misanthropic, was published November 18, 2026 by the Center for AI Safety (CAIS), which happened on the eve of the release of the Claude developer on the Nasdaq exchange. Signatories of the appeal include economist Oliver Hart, physicist Sheldon Glashow, artificial intelligence specialist Judea Pearl, Hugging Face co-founder Thomas Wolf, and xAI co-founder Igor Babushkin. The authors of the appeal accuse Anthropic of using its reputation for AI safety to attract investors and employees, while simultaneously retreating from its previously stated commitments.

One of the main points of criticism is Anthropic’s work on AI recursive self-improvement. We are talking about creating systems that can participate in the development and training of their more advanced successors. Critics point to a contradiction between Anthropic’s previous warnings about the risks of such a process and the company’s own use of AI to speed up the development of new models. In registration materials filed with the US Securities and Exchange Commission (SEC), Anthropic describes a cycle in which each new version of Claude helps create the next. According to the data provided, already in September 2026, Claude accounted for more than 90% of the contribution to writing program code at the request of the company.

Another complaint concerns the revision of the so-called responsible scaling policy (RSP). In February 2026, Anthropic changed this document, abandoning its previous unconditional commitment to independently suspend the development of models if the company fails to provide sufficient protection measures. In other words, if controls do not keep pace with the development of AI.

The signatories also accuse Anthropic of lobbying for legislative initiatives in California and New York aimed at regulating AI. According to them, the company is seeking to ease restrictions that could affect its own activities. In a statement released, CAIS called on Anthropic to stop using its security-focused reputation to justify concentrating AI development efforts in virtually one or more hands.

According to the signatories, Anthropic is trading on its “safety first” reputation to win over investors, employees and the general public, while quietly backtracking on previous commitments. The authors of the appeal openly call Anthropic “an unsafe artificial intelligence company.”

It is interesting to note that this call was signed by both opponents of artificial intelligence, demanding to stop its development altogether, and fighters for its further improvement. Anthropic’s stock debut brought together previously irreconcilable parties. The former, in principle, do not trust AI, while the latter are afraid of the concentration of AI in one hand.

Alas, the appeal did not help reduce the financial interest of the securities market in Anthropic shares. According to the SEC, on November 18 the company set the offering price at $83 per share, offering investors 980 million shares. The IPO volume was about $81.34 billion, and the company’s estimated capitalization at the time of placement reached approximately $1.5 trillion.

Additionally, Anthropic raised $26 billion through a private placement. Among the investors named are AMD, Broadcom, Google, Micron, Microsoft, Nvidia and TSMC. Trading on Nasdaq under the ticker ANT began on November 19. On the first day, shares opened at $111.50, 34% above the offering price, and the company’s market valuation briefly reached approximately $2 trillion and continues to grow, setting all possible records.

If you notice an error, select it with the mouse and press CTRL+ENTER.