Anthropic's Claude AI identifies novel CRISPR‑like enzyme system

Anthropic, the U.S.-based artificial‑intelligence firm, announced on Saturday that its large‑language model Claude identified a previously unknown enzyme system that functions similarly to the CRISPR gene‑editing machinery, according to Al Jazeera.

The AI model scanned millions of protein sequences from public databases and flagged a set of proteins that share structural motifs with Cas nucleases, the molecular scissors used in CRISPR. Subsequent laboratory validation by Anthropic’s partner labs confirmed that the enzyme can cut DNA at targeted sites, opening a potential new toolkit for genetic engineering.

Anthropic’s chief scientist described the find as “a proof‑of‑concept that generative AI can accelerate basic biology research,” Al Jazeera reported. The company said the discovery could shorten the timeline for developing disease‑targeting therapies, though it emphasized that commercial applications are still years away.

CRISPR, first harnessed in 2012, revolutionised molecular biology by providing a relatively simple, cheap method to edit genes. Since then, dozens of variants have been engineered to improve precision, reduce off‑target effects, and expand the range of organisms that can be edited. A novel enzyme with CRISPR‑like activity adds to this expanding toolbox and may enable edits that current systems cannot achieve.

The breakthrough arrives amid a growing debate over how to govern AI‑driven biotechnology. Experts warn that the same models capable of suggesting new enzymes could also be misused to design harmful pathogens. International bodies such as the WHO and the UN have called for frameworks that balance innovation with bio‑security, and several governments are drafting regulations on AI‑generated biological data.

Anthropic said it will share the enzyme’s sequence with the scientific community under controlled access, and it plans to collaborate with academic groups to explore therapeutic uses. The company also announced internal safeguards, including a review board that evaluates any AI‑suggested biological discovery for dual‑use risk before publication.