AI Giant Anthropic's Biology Lab Stumbles Upon Major Breakthrough!
Anthropic's new wet biology lab, powered by its AI model Claude, has reportedly discovered a novel enzyme system with CRISPR-like properties. This finding highlights the increasing role of AI in scientific discovery, though ethical concerns regarding AI's capabilities and the human-led nature of current lab experiments remain central to the discussion.
Last week, AI giant Anthropic confirmed the operation of a wet biology lab in the Bay Area, utilizing its advanced AI models to conduct physical experiments. This week, the company announced what it believes to be a significant discovery from this lab: a novel enzyme system possessing properties reminiscent of CRISPR, as described by Anthropic. CRISPR, a natural immune system found in bacteria to combat viruses, has become a widely adopted gene-editing technology in research.
Anthropic's researchers claim to have unearthed a previously unknown enzyme system within the DNA of bacteriophages – viruses that infect and replicate within bacteria. This system reportedly mimics CRISPR's capabilities, capable of performing operations such as cutting, copying, and pasting DNA. While the broader research community is tasked with validating the novelty and significance of this discovery, Anthropic CEO Dario Amodei acknowledged that the finding built upon prior work, referencing a similar system previously discovered by a Stanford team.
A notable aspect of this announcement is Anthropic's emphasis on Claude, their AI, being primarily responsible for the discovery. Amodei stated that the enzyme system was found "mostly, though not entirely, by Claude." The lab, established only this spring, made this discovery rapidly, with Claude reportedly completing the concerted effort in just 21 hours. During this period, Claude searched through data using approximately 950 agents and processed 210 million tokens.
This revelation emerges shortly after AI CEOs, including Amodei, publicly addressed the growing capabilities and potential dangers of AI models, advocating for a slowdown in development and the implementation of robust safety-testing procedures. This follows warnings from Anthropic employees about the potential for AI to pose existential risks. Despite Amodei's personal fears regarding AI's misuse for bioterrorism, he also harbors optimism that AI will "cure most diseases in 5-10 years," indicating Anthropic's decision that the potential rewards outweigh the risks.
Crucially, the physical experiments at Anthropic's lab are currently performed by human scientists, not fully by AI. The lab operates at lower biosafety risk levels (BSL-1 and BSL-2) and does not handle human-infecting pathogens. Anthropic's work fits within a broader trend of AI-powered biological research; Stanford researchers are collaborating with LLMs and CRISPR, UC San Francisco has used AI to design enzymes, and Google launched its AI-powered AlphaFold tool in 2020. However, Amodei has not dismissed the possibility of a fully automated biology lab in the future, envisioning Claude autonomously controlling lab equipment with appropriate safeguards.