Anthropic's ART finding looks more like genomic search than discovery

What Anthropic reported

The agents found recurring DNA sequences next to the reverse transcriptase and grouped them into a system called ART. The sequences occur mainly in bacteriophages, viruses that infect bacteria.

The company has not established ART’s function. At this stage, the result is a genomic pattern and a promising lead, not a demonstrated biochemical process.

The comparison with CRISPR

The strongest criticism came from Harrington, who earned his PhD under Nobel laureate Jennifer Doudna and co-founded Mammoth Biosciences with her. Mammoth itself uses genomic searches to identify CRISPR systems.

On X, Harrington wrote that this kind of method has existed for decades and that similar systems have been known since at least 2008. He argued that the difficult part is determining what a system actually does—something Anthropic has not yet shown.

That distinction is the entire story. Finding a repeated sequence is not the same as finding a function.

What the announcement leaves open

My read is that Anthropic has presented an interesting search result as the beginning of a biological discovery, while the evidence currently supports the narrower claim. The unusual reverse transcriptase and the ART sequences may point toward something important, but the announcement does not establish activity, mechanism, or practical use.

The quiet part is the experimental one: no function has been demonstrated. Until that happens, ART is better understood as a candidate system surfaced by AI-assisted genomic search than as a new enzyme discovery.