Amodei trims AI-biotech hype; FDA clears Trutakna
Get the Health newsletter
Daily health & science — research, biotech, public health, the studies worth knowing. Free.
- FDA approved Vera Therapeutics' Trutakna on Tuesday for IgA nephropathy, a chronic autoimmune kidney disease with no cure that can progress to dialysis.link ›
- Dario Amodei told STAT that his 'century in a decade' AI-biotech forecast from his 2024 essay 'Machines of Loving Grace' isn't visible for at least a decade, citing model capability gaps, researcher adoption lag, and infrastructure/regulatory delays.link ›
- Anthropic launched Claude Science for biologists and pharma companies at the same event where Amodei conceded the AI-biotech timeline slip.link ›
- Cardiff University senior lecturer Simon Scofield and team found a 6% catnip oil lotion matched 15% Deet in eastern Uganda field trials, with a 2% concentration only marginally less effective.link ›
- Diagnostic stewardship at 46 Michigan hospitals — requiring true UTI symptoms before ordering a urine culture — cut unnecessary antibiotic treatment from roughly 29% to under 17%.link ›
- Naoki Morimoto at Kyoto University led the first full dig at Turkey's Üçağızlı II cave since 2020, recovering nearly 20,000 stone artefacts and ~30 decorative Columbella rustica shells suggesting Neanderthal–Homo sapiens contact over 77,000–47,000 years ago.link ›
- Katerina Harvati at Tübingen and Mark Hubbe at Tennessee analyzed 87 hominin skulls and found neutral drift — not natural selection — best explained braincase growth across Homo species.link ›
The most honest biotech moment of the week came from Anthropic's Dario Amodei, who told STAT that his 'century in a decade' forecast for AI-driven life sciences isn't visible for at least a decade — a rare CEO walkback grounded in model capability gaps, researcher adoption lag, and infrastructure/regulatory friction. The FDA also approved Vera Therapeutics' Trutakna for IgA nephropathy on Tuesday, giving patients with the chronic autoimmune kidney disease — which has no cure and can progress to dialysis — a newly approved targeted therapy. Anthropic simultaneously launched Claude Science for biologists and pharma. The Trutakna win is concrete; Amodei's admission is the real story — biotech's loudest AI preacher publicly conceding the disruption dividend is further out than the loudest voices claimed.
The stories behind this week
FDA Approves Vera's Trutakna for IgA NephropathyIgA nephropathy patients gain a newly FDA-approved treatment for a progressive autoimmune disease that can advance to organ failure and dialysis, while Vera Therapeutics clears a key regulatory milestone for its lead drug. The approval opens a commercial pathway for a condition with no cure and limited treatment options.
Catnip Lotion Matches Deet in Uganda TrialsFor rural Ugandan subsistence farmers who can't afford commercial repellents, a locally producible catnip lotion matching Deet's effectiveness at a fraction of the cost could become a new tool against malaria, which killed 610,000 people in 2024. But an independent expert's compliance caveat means it would supplement — not replace — existing vector-control measures like insecticide-treated nets.
Amodei Walks Back 'Century in Decade' AI-Biotech ClaimAmodei — one of the most vocal AI CEOs about life-sciences transformation — is publicly dialing back his own hype, which matters for biotech and pharma R&D leaders who have been planning around compressed timelines. The source attests he cited model capability gaps, researcher adoption lag, and infrastructure/regulatory delays as the three concrete barriers. Anthropic simultaneously launched Claude Science, signaling commercial intent even as the timeline slipped.
Opinion: The AI licensure debate is missing the point of licensureThe authors — who lead AI policy at the SCAI, the American College of Cardiology, and the FDA's Digital Health Advisory Committee — are directly shaping how AI enters cardiology. Their argument carries weight because the legal duty of care, affirmed across four cited cases, cannot be transferred to software, meaning every hospital deploying AI diagnostics must keep a licensed physician as the accountable decision-maker or absorb the liability themselves.
The strange metals forcing us to rethink how electricity really worksThe entire electronics industry rests on Lev Landau's 1950s quasiparticle framework. If strange metals' resistance is driven by collective electron patterns rather than individual particle collisions — as the 2025 neutron data, Sachdev's model, and holographic approaches together suggest — that would force a rewrite of fundamental conduction physics and could finally crack high-temperature superconductivity, the 1980s breakthrough whose promised payoff of lossless power transmission remains unrealized.
Turkey cave artefacts hint Neanderthal-human exchangeFor decades, Neanderthals were framed as cognitively inferior to modern humans. The shared collection of aesthetically valued but non-utilitarian shells suggests cognitive parity and possible cultural transmission between species, reshaping how researchers model early human migrations and the symbolic capabilities of Neanderthals.
Human brain growth wasn't driven by natural selectionThis challenges the textbook 'brains grew because being clever was advantageous' narrative central to paleoanthropology: if neutral drift drove brain expansion, the cognitive capabilities defining Homo sapiens may be partly a byproduct of constraints releasing (possibly via cooking) rather than direct selection for intelligence, reshaping how researchers interpret what makes humans unique.
Opinion: Stopping doctors from ordering unnecessary diagnostic tests requires a structural fixStructural changes to test ordering cut inappropriate antibiotic use from 29% to under 17% across 46 Michigan hospitals without new drugs, pointing to 14 billion annual U.S. tests as a major cost and harm-reduction target. The CDC has called for stewardship programs but lacks a confirmed director to enforce them.
Why it matters: Amodei publicly trimming his 'century in a decade' AI-biotech timeline — while Anthropic ships Claude Science anyway — forces pharma R&D leaders to rebase AI-disruption roadmaps that until now were built on a decade-scale compression even the loudest AI CEO now concedes is unrealistic.
