Anthropic Expands AI Ambitions into Drug Science

SAN FRANCISCO, Sept 18 – Anthropic has quietly established a laboratory dedicated to physical biology, marking its entry into drug science.

With public anxiety surrounding AI on the rise, the startup has developed a wet lab in the San Francisco Bay Area, according to two sources familiar with the situation.

Anthropic has expressed intentions to create treatments for rare diseases, extending its efforts beyond mere computer-based evaluations.

During a recent interview with Reuters, Eric Kauderer-Abrams, Anthropic’s head of life sciences, confirmed the existence of the new lab.

“We believe that to do biology, the final test is still and will be for a while in real lab work,” he stated.

Kauderer-Abrams also described the company’s approach as typical of many biotech firms, combining in-house experiments with collaborations with external partners.

Notably, a spokesperson later clarified that the lab isn’t specifically aimed at drug discovery, though they did not elaborate further.

This initiative represents a small step towards Anthropic’s larger goal of addressing diseases overlooked by the pharmaceutical industry—doing so at a time when there’s a pressing concern about AI displacing jobs and affecting human lives, as one source indicated.

However, the success of any drug programs is uncertain since most drugs tend to fail during clinical trials for safety and efficacy.

For Dario Amodei, the CEO of Anthropic, this mission is personal. He shared that a disease claimed his father’s life just before a cure was discovered. Kauderer-Abrams echoed this sentiment, saying, “The mission of the company is to develop powerful AI in such a way that benefits the world.” He added, “By far, we see the biggest opportunity for that in the life sciences, and that is motivating everything that we’re doing.”

Balancing Ambition with Fears of Catastrophe

Similar to other biotech firms, Anthropic is embracing physical automation, as per insights from the two unnamed sources.

The startup aims to enhance how its Claude AI commands robotic systems to conduct scientific experiments with minimal human intervention, according to one individual.

Nonetheless, the company maintains that human oversight is crucial for safety, as emphasized by its spokesperson.

“We’re in the very early innings of using AI to automate the execution of lab work,” Kauderer-Abrams remarked, noting the potential for significant acceleration across various processes.

Anthropic’s endeavor to integrate rapidly evolving technology into a biology lab comes at a pivotal moment for the organization.

Recently, some researchers within the company expressed concerns that AI could potentially threaten human existence. Amid this backdrop, Anthropic noted instances where its technology could have been misused for biological weapons development, which drew skepticism.

Amodei has also mentioned the need for a slowdown in AI development, warning that an advanced AI could cause catastrophic damage.

Meanwhile, Anthropic is gearing up for a public offering that could raise around $2 trillion to further its AI initiatives.

Kauderer-Abrams said, “In our life-science work, we’re always balancing these things—making models better in ways that we think will uplift and accelerate developing medicines while being careful about limiting risks.”

Procurement Ramp-Up

Anthropic made its intentions clear to begin running drug programs during an event in San Francisco this past June. Kauderer-Abrams announced that the company would carry out preclinical research in areas considered financially unattractive by traditional firms.

The startup has also launched software named Claude Science and welcomed Novartis CEO Vas Narasimhan to its board, in addition to acquiring Coefficient Bio for about $400 million in stock, according to media reports.

While Anthropic confirmed the acquisition, it refrained from commenting on the financial specifics.

Currently, the company is enhancing its laboratory efforts to gain firsthand experience in biology, though it still outsources some work when it proves more efficient, Kauderer-Abrams noted.

“There are certain tasks we can execute much faster ourselves,” he stated, aiming for the largest possible scale in operations.

Anthropic is seeking to hire a leader to enhance procurement and other operations, as highlighted in a recent job posting. Another position advertised calls for expertise in “protein and nucleic acid characterization,” along with other biochemical responsibilities. One job listing emphasizes, “Our goal is to speed up progress in the life sciences by an order of magnitude.”

Kauderer-Abrams indicated that life sciences represent one of Anthropic’s largest areas of investment in terms of personnel and resources.

Targeting ‘Undruggable’ Conditions

He believes that AI could expedite the development of conditions that were previously deemed “undruggable,” which is to say, too challenging to treat.

The technology may also accelerate the discovery of complex molecules, such as bispecific and trispecific antibodies, which can target multiple locations on a specific protein or cell.

However, the exact diseases that Anthropic is focusing on and its progress remain somewhat vague. The journey from identifying a molecule to bringing a drug to market can take years, particularly as it navigates human trials, a challenge that Anthropic has not yet undertaken.

By comparison, Isomorphic Labs, a drug discovery branch of Alphabet, has been pursuing clinical pathways for several years, though it recently pushed its goal back to 2026.

In its pursuit of laboratory and automation advancements, Anthropic unveiled the Model Hardware Standard in August, designed to optimize AI-operated equipment. A separate academic initiative called PyLabRobot has also addressed automation needs, albeit in a different context.

Anthropic provides tools and AI services, including the Model Hardware Standard, to leading pharmaceutical companies like Roche, Bristol Myers Squibb, and Novo Nordisk.

At the same time, Anthropic is grappling with a trust issue. Sources mentioned that potential customers are concerned about the possibility of Anthropic learning from their competing drug development efforts, even though the company assures that data security measures are in place.

Kauderer-Abrams acknowledged that, partly due to competitive concerns, Anthropic has decided to steer clear of running clinical trials for now, directing focus toward areas the industry typically neglects.

“We’re not competing with pharma and biotech companies that make their business in bringing drugs to market,” he clarified.