Dario Amodei will appear before the United Nations Security Council on Wednesday, not in person but over a video link, to tell its members about the risks of artificial intelligence. It will be the first time the leader of a frontier AI lab has briefed the body, and it comes at an unusual moment for his company.
Anthropic, the startup Amodei co-founded and runs, is doing three things at once. It is warning governments about the dangers of AI while preparing to sell shares to the public, and it has quietly built a biology laboratory in the San Francisco Bay Area where it is testing whether its Claude model can direct robots to run experiments.
The Security Council briefing, reported by Bloomberg on September 22, is scheduled for September 23 and will be chaired by France. Amodei will speak alongside Sam Altman of OpenAI, the researcher Yoshua Bengio and Hugging Face’s Clement Delangue. The session is expected to produce no binding text, according to people familiar with the plans.
The same day, the New York Times reported that Anthropic is pushing ahead with plans for an initial public offering even as it argues in public for caution about the very technology it sells. The two stories, appearing within hours of each other, capture a company that is both a leading critic of unfettered AI development and a business racing to capitalize on it.
Anthropic was founded in 2021 by Dario Amodei and his sister Daniela, both former OpenAI executives who left over disagreements about safety and direction. The company has since raised billions from backers including Amazon and Google to build Claude, a family of models that competes with OpenAI’s GPT for enterprise customers.
The wet lab is the newest piece. Reuters reported that Anthropic has set up the laboratory to move beyond the computer simulations that have defined its biology work, and that it wants to push Claude to run robotic experiments with limited human supervision. Eric Kauderer-Abrams, the company’s head of life sciences, confirmed the lab, while a spokesperson said it is not specifically for drug discovery.
The lab work is part of a broader push into drug science. Anthropic launched a software product called Claude Science, and on September 16 it announced a partnership with Novo Nordisk, the Danish drugmaker behind Wegovy, to apply Claude to the search for new medicines. Novo said it would test Claude Science on research and development workflows and use Anthropic’s models to speed up its own software engineering.
Amodei framed the partnership in sweeping terms, saying AI’s growing capability could “compress a century’s worth of biological and medical breakthroughs into a decade.” Anthropic has also added Novartis chief executive Vas Narasimhan to its board and bought the startup Coefficient Bio for about $400 million in stock, according to media reports.
At a San Francisco event in June, Kauderer-Abrams said Anthropic would run preclinical work in disease areas that traditional drugmakers avoid because they are not attractive enough financially, and he has pointed to rare diseases and molecules once thought undruggable as the kind of targets AI could unlock. Anthropic has said it will stop short of clinical trials, a line that keeps it from competing directly with the drugmakers it now counts as partners.
The company has been assembling the pieces for that work for more than a year. It acquired Coefficient Bio to build tools for drug development, launched a standard to help AI operate laboratory equipment, and added Narasimhan to its board. Rival Alphabet has a similar effort in Isomorphic Labs, which recently delayed its own push toward the clinic to the end of 2026, evidence of how long the road from a promising molecule to a tested drug can be.
All of this sits awkwardly alongside Amodei’s public warnings. On September 12 he published an essay urging the field to apply the brakes at the frontier, arguing that the industry is moving too fast to manage the risks. Less than two weeks later, the company’s expansion into a physical lab, a public listing and a Security Council appearance were all moving forward at once.
Anthropic’s backers do not seem troubled by the tension. The company has raised billions at valuations that place it among the most valuable private companies in the world, and its IPO preparations suggest investors read the warnings as a feature rather than a contradiction, a sign that the company takes the technology seriously.
Analysts said the strategy amounts to a bet that a company can be both the industry’s conscience and one of its largest commercial players. Whether that holds will be tested as Anthropic moves toward a public listing, where it will face quarterly pressure to grow revenue even as it argues for restraint.
The wet lab illustrates the ambition. A company that began by trying to make AI systems safe is now using those systems to run physical experiments, with an eye toward drugs once thought impossible to develop. If it works, the payoff could be enormous; if it raises new safety questions, Anthropic will be answering them from a laboratory it built itself.
For now, the company is moving on all three fronts at once, and the Security Council will get a window into that on Wednesday, when its members listen to a man who is at once warning about the technology and building a business on top of it.
