We took the summer off, and we regret nothing. There were lakes. There was a sauna. There was an unreasonable volume of new potatoes. Meanwhile, biotech kept working through July like an intern trying to get noticed, so we came back to a pile of announcements all wearing the word "first" and most of them needing an asterisk. Let's sort the actual news from the press releases.
Table of Contents
🫧 Midjourney Built a Body Scanner. It Is Not an MRI.
NEWS
The company known for generating pictures of dragons in Renaissance armor now wants to generate pictures of your pancreas. In June, Midjourney unveiled a whole-body scanner, and roughly every headline called it an MRI. It is not one. No magnets, no liquid helium. There is, instead, a shallow pool of water.
You stand on a platform that lowers you through a ring packed with 358,000 ultrasonic elements firing up to a thousand times per second, essentially echolocating you like a very expensive dolphin. It is ultrasound computed tomography, built on imaging chips licensed from Butterfly Network in a deal worth up to $74 million over five years. Butterfly's stock leapt 33% on the news, because of course it did.
The pitch is a 60-second scan at a spa. The prototype takes about 20 minutes, has been used on about a dozen people, and runs without any AI in its imaging pipeline. From the AI company. It has no FDA clearance to diagnose anything, so it launches as wellness-flavored "body composition maps."
Then there is the claim that enough early imaging could help the world avoid 30% of all deaths and 50% of all healthcare costs, a number with no peer-reviewed support. Radiologists also note the physics: Johns Hopkins' Francis Deng flags the inability of ultrasound to penetrate bone, air, and deep soft tissues. And screening healthy people at scale reliably produces false positives and incidental findings that send people for unnecessary, sometimes invasive follow-ups.
To be fair, the hardware is real engineering. But the actual accessible-imaging story is Hyperfine's portable brain scanner, which runs at 0.064 tesla and has been FDA-cleared since 2020, with no golden light and no spa.
🧠 The First Tau Gene Therapy Gets Permission to Exist
NEWS
Alzheimer's drug development has spent two decades mostly poking at amyloid plaques and producing results best described as "statistically detectable." Tau, the protein that tangles inside neurons and tracks much more closely with cognitive decline, has been the harder target. Voyager Therapeutics just got the regulatory green light to go after it with gene therapy.
On June 1, the FDA cleared Voyager's Investigational New Drug application for VY1706, which the company says is the first such clearance for a tau-targeted gene therapy. Important translation: an IND clearance is permission to begin a trial in humans. It is not an approval, and nobody has been dosed yet.
The design is the interesting part. It is a vectorized siRNA that targets MAPT, the gene encoding tau, delivered by an engineered capsid that crosses the blood-brain barrier using the ALPL receptor. One intravenous infusion, no brain surgery, and the capsid is deliberately built to skip the liver. In July, Voyager reported that a single dose achieved up to 75% lowering of MAPT mRNA and tau protein in key Alzheimer's-relevant brain regions through six months.
Restraint required: that 75% is in monkeys, from a toxicology study. The human trial is an open-label dose escalation in up to 18 adults with early Alzheimer's, with dosing expected in the second half of 2026. Eighteen patients cannot tell you whether anyone thinks more clearly. They can tell you whether a one-time injection safely lowers tau in a living human brain.
Which, honestly, would be enough of a headline on its own.
🧬 Someone Base-Edited Human Embryos, and the Funding Is the Scary Part
RESEARCH
Every few years a lab edits human embryos and the field collectively updates its anxiety. This time the science is incremental, and the money is not.
Dieter Egli's team at Columbia used base editing, which nicks a single strand of DNA and swaps a letter instead of cutting the double helix, on human embryos at two targets, PCSK9 and the fetal hemoglobin genes. The embryos were never implanted, no pregnancy was intended, and the work sits on bioRxiv as a preprint that has not been peer reviewed. Delivering the editor as protein allowed normal development to the blastocyst stage. Delivering it as RNA arrested the embryos entirely. Most edited embryos were mosaic, meaning different cells ended up with different genomes.
Note what was not done: nothing was corrected. As Science reported, the team inserted mutations rather than fixing any, and the New York Times quietly dropped "first" from its headline the next day. Chinese labs have been base-editing embryos since 2017. The novelty here is the venture capital.
Which is where the reaction gets loud. Alexis Komor of UC San Diego told Scientific American the work "kind of opens the floodgates," calling it a gateway to editing for enhancement.
The uncomfortable detail is the backing. Nucleus Genomics, an embryo-screening startup whose CEO is a Thiel Fellow and whose investors include Peter Thiel's Founders Fund, is financing the next stage of Egli's work. The Center for Genetics and Society's Katie Hasson responded that germline decisions "certainly should not be driven by Silicon Valley's transhumanist fantasies."
💉 An AI Designed a Vaccine, and Humans Took It
NEWS & RESEARCH
"First AI-designed vaccine injected into people" is a great headline. The paper underneath it is more of a polite shrug, and that gap is the story.
A Cambridge team and its spinout DIOSynVax built pEVAC-PS, a DNA vaccine whose antigen was designed computationally rather than copied from a real virus. The machine learning chewed through sarbecovirus sequence databases to construct a synthetic super-antigen stitched from the bits that stay conserved while coronaviruses mutate. Cambridge states this is the first time a vaccine whose active component was designed entirely by computer simulations has been tested in humans. Delivered needle-free, by jet, straight into the skin.
Then the results. A total of 39 volunteers were vaccinated between December 2021 and September 2023, so the news is really the publication catching up with an old study. The vaccine was safe and well tolerated, which is what a phase I asks. Immunogenicity was another matter: antibody responses were modest and variable, with only small increases against Delta and Omicron and limited activity against the ancestral Wuhan strain and SARS-CoV-1.
Outside experts were kind but firm. Marc Boubnovski of Novo Nordisk noted it did not yet show the strong, broad immune response you would want before calling it a protective universal coronavirus vaccine. It is also worth knowing that multiple authors are employees or shareholders of DIOSynVax, which they disclosed properly.
So: the computer designed a protein, the protein was safe, and biology has not voted yet.
🪰 A Second Fly Has Been Cleared to Eat Your Dead Tissue
NEWS
Modern wound care has lasers, growth factors, hyperbaric oxygen, and engineered skin substitutes. It also has larvae, which remain undefeated at the specific job of eating dead tissue and leaving healthy tissue alone.
Cuprina Holdings announced on June 15 that the FDA had cleared MEDIFLY Maggots, the first US clearance for a maggot debridement product using the Lucilia cuprina species, better known as the Australian sheep blowfly. One correction to the coverage: this is not a newly discovered species. It is a newly cleared one, which is a regulatory milestone rather than a taxonomic one.
The mechanism is properly gross and properly clever. Larvae secrete proteolytic enzymes and antimicrobial peptides that liquefy slough and disrupt bacterial biofilms, the films that make chronic wounds shrug off antibiotics. The clearance came through on substantial equivalence to the green bottle fly product cleared back in 2004, a clearance held by Dr. Ronald Sherman, who is now Cuprina's medical and scientific director. CEO David Quek called holding both species a position "no other company holds." A competitive moat made of flies.
The stakes justify the squeamishness. People with diabetes face a 19% to 34% lifetime risk of a foot ulcer, and the US performs roughly 150,000 diabetes-related amputations a year. Maggots clear dead tissue faster than conventional treatment, though the trial evidence has not shown a clear improvement in overall healing.
A supply chain, in this case, is a fly farm.
Every story this week arrived wearing the word "first," and every one of them needed a footnote. The MRI was an ultrasound. The gene therapy has permission rather than results. The embryo editing was neither the first nor a treatment. The AI vaccine cleared safety and not much else. The maggot is older than the FDA.
None of that makes them boring. A one-time infusion that silences tau in a human brain would be extraordinary. A second maggot species means redundancy for people whose alternative is amputation. The science is doing fine. It is the adjectives that need supervision.
Which of these made you say "wait, what" out loud? Reply and tell us, especially if you are a radiologist with opinions about spa-based screening. And if a friend has been forwarding you Midjourney MRI headlines all summer, send them this instead.
Keep questioning everything (especially the word "first"),
P.S. Hyvää rapukautta to our Finnish readers. Between the crayfish and the maggots, this newsletter has become unexpectedly invertebrate-heavy, and we are choosing to see that as range.