Opinion · 2 min read

Stop Pricing ART Like CRISPR Until Enzyme Activity Assays Land

Classy view: Anthropic deserves credit for ART sequence discovery, but CRISPR comparisons and UN stage hype outrun the preprint's missing functional assays.

By Classy AI News · September 23, 2026

Stop Pricing ART Like CRISPR Until Enzyme Activity Assays Land

What changed

Anthropic’s 23 September 2026 ART enzyme announcement triggered immediate CRISPR analogies in executive remarks and headlines. At the UN Security Council the same day, chief executive Dario Amodei referenced the biology result while urging safeguards, a sequence observers including Gary Marcus noted as premature hype relative to the underlying evidence. The company’s preprint explicitly states researchers have not shown that the reverse transcriptase is active or that array RNAs are substrates, even though early expression data exist.

Researcher examining cell cultures under a sterile hood in a biotech lab
Figure: Repeat arrays invite CRISPR comparisons long before editing tools are validated.

Virologist Angela Rasmussen, quoted in post briefing commentary, said Anthropic found the sequence, probably expressed a protein, but have not done any meaningful functional characterization, adding it is very premature to compare to CRISPR. Classy agrees: discovery credit is due, platform credit is not.

Why it matters

Biotech investing already chases CRISPR shaped multiples. When frontier labs pair autonomous discovery claims with geopolitical testimony on the same news day, allocators risk pricing optionality that assays have not confirmed. That distortion spreads into partnership terms for pharma AI vendors who actually deliver validated assays, not just genomic search throughput.

Financial analysts reviewing biotech investment charts on large monitors
Figure: Markets reward editing narratives faster than labs produce mechanism papers.

Who is affected

Venture partners, corporate development teams, and science communications leads at AI biology startups should tighten claim review. Limited partners should demand milestone tables tied to enzymatic activity, not agent hours searched. Regulators watching autonomous genomic agents need clarity that discovery automation does not shortcut biosafety tiers.

What to do next

Adopt a public claims policy: repeat architecture parallels to CRISPR require independent functional editing or cleavage data before investor decks use platform language. Internally, separate lead identification KPIs from tool validation KPIs in every AI biology contract.

What to watch

Peer reviewed ART mechanism studies. Whether Anthropic publishes assay progress on RT activity. Competitor attempts to replicate the search harness on public metagenomic sets without matching hype in press cycles.

Sources

  1. Primary. Anthropic preprint PDF, Autonomous AI agents discover reverse transcriptases with tandem repeat arrays (2026). States limitations on RT activity and substrates.
  2. Secondary. Gary Marcus Substack, Historic UN Security Council Briefing on AI (23 September 2026). Documents Amodei CRISPR comparison timing and quotes Angela Rasmussen on premature characterization.
  3. Secondary. The Next Web, Anthropic says Claude found a new enzyme system with CRISPR like repeats (23 September 2026). Summarizes array structure and unproven enzymatic activity.

Newsletter

Get the dispatch

One field. One email when we publish. Privacy.