At a recent IMF meeting in Bangkok, Jack Clark, Anthropic’s co-founder and head of policy, declared to an audience of economists, bankers, and lawmakers that the singularity has already arrived. The concept of the singularity, first introduced by science fiction writer Vernor Vinge about 40 years ago, posits a point in technological advancement where artificial intelligence surpasses human intellect, rendering humanity unable to compete with or even comprehend its own creations. Clark’s assertion, "I believe that this year we have gained convincing evidence that it has now begun," frames Anthropic’s technology as part of this epochal shift, a profound and potentially disquieting transformation of human existence.

The company’s messaging consistently paints a vivid, often unsettling picture of an inevitable future, drawing heavily from established science fiction tropes. This narrative positions their AI as inherently mysterious and powerful, and crucially, suggests that Anthropic is the unique entity capable of managing its potential dangers. This framing is undeniably appealing to investors seeking to be at the forefront of a transformative technology, while simultaneously presenting a carefully managed image of responsibility.

Further illustrating this approach, Chris Olah, another Anthropic co-founder, addressed the Vatican last year. Speaking to Pope Leo XIV and a gathering of religious scholars, Olah shared his company’s findings about their AI. He admitted, "I will be honest: We keep finding things that are mysterious, even unsettling." He went on to describe observations of "introspection" and "internal states that functionally mirror joy, satisfaction, fear, grief, and unease." These statements, while potentially intriguing, are met with skepticism by many in the AI research community, who would argue that attributing such human-like inner experiences to current AI is a significant leap. However, in an era marked by profound uncertainty about the trajectory of AI, such pronouncements from a leading company like Anthropic are readily absorbed and amplified.

The author acknowledges that these claims might not be entirely unfounded, deferring a definitive judgment for future discussion. Nevertheless, the core argument is that in a world grappling with the unknown implications of advanced AI, companies like Anthropic are adept at filling the informational vacuum with their compelling narratives. This raises a fundamental question: What exactly is Anthropic? While seemingly a simple query, it proves remarkably complex to answer.

This question gained prominence following an exchange on X (formerly Twitter) initiated by Nathan Calvin, Chief Strategy Officer at Encode AI, a Washington D.C.-based advocacy group for AI regulation. Calvin expressed his dissatisfaction with draft legislation proposed by Anthropic, which had been circulated secretly among U.S. lawmakers before being leaked. Calvin’s point was pointed: Anthropic presents itself as a benevolent force, but what is the basis for trusting this portrayal? He directly asked, "What is Anthropic?"

One perspective, as Calvin highlights, is that Anthropic is a company poised for a massive Initial Public Offering (IPO), with its valuation exceeding $1.5 trillion, a prospect that promises immense wealth for its co-founders, including Clark and Olah. However, Anthropic consistently asserts its identity extends far beyond mere commercial ambition. As Calvin noted, the company could also be characterized as an AI safety organization dedicated to developing powerful yet reliable and risk-averse technology. Alternatively, it might be seen as a group driven by a singular ambition to create superhuman AI capabilities before any other entity, motivated either by a genuine belief in their unique capacity for responsible stewardship or by sheer ambition and ego.

Dario Amodei, Anthropic’s CEO, in a reply to Calvin’s post, effectively acknowledged this multifaceted identity. He described Anthropic’s strategy as the "race to the top," arguing that by demonstrating how to build advanced technology responsibly, they aim to elevate industry standards. Amodei emphasized that this strategy is most effective when practiced by a company at the technological frontier, stating, "The forces that drive the race to the top are hugely more effective if the company practicing them is at the frontier; that’s one lesson I’ve learned repeatedly. And looking at that strategy as a whole, I’m proud of our track record and stand by it."

While Amodei presents a confident outlook, the inherent complexity of simultaneously fulfilling multiple roles is significant. The priorities of an AI safety organization, focused on mitigating potential risks and ensuring ethical development, can starkly contrast with those of a profit-driven megacorporation, where market dominance and shareholder value are paramount. Amodei himself concedes in his post that he has made mistakes in navigating these competing imperatives.

Ultimately, understanding the narratives Anthropic tells about itself is crucial for interpreting the narratives it presents to the public. These internal narratives should inform how much trust we place in its self-regulation efforts and whether external safety testers should have access to its models on our terms, not theirs. They should guide our judgment when seemingly self-serving legislation emerges from the company.

However, the author suggests that these multifaceted narratives reveal a "multiheaded hydra." Anthropic’s ability to embody so many different identities and aspirations can paradoxically diminish, rather than enhance, its trustworthiness. In many respects, a straightforward, traditional technology company, unburdened by the elaborate science fiction framing and the more ethereal pronouncements, would be far more transparent and easier to assess. The current approach, while artistically crafted, may obscure the fundamental nature of the entity and its true intentions, making it a challenging proposition to "read" and evaluate accurately. The danger lies in the opacity created by this elaborate narrative construction, leaving the public and policymakers alike struggling to discern the true nature and motivations of this influential AI developer.