🧬 AI Is Coming for Your Pathogens (And Also Your Vocal Cords)

Welcome to another week where biotech decided that science fiction is just non-fiction with better marketing. We've got OpenAI trying to prevent AI-designed bioweapons (because of course someone needs to), scientists printing tissue inside your throat during surgery (casual), and the last two northern white rhinos getting some serious reproductive tech support. Also, there's a new malaria drug that works great except for the part where it tastes so bad people vomit, and Eli Lilly just reminded everyone that GLP-1 drugs aren't the only way to lose 20% of your body weight. Turns out biology has multiple cheat codes, and we're discovering them faster than we can figure out what to do with them. This is fine.

RESEARCH & NEWS

Valthos Inc. emerged from stealth in October 2025 with $30 million and a mission statement that reads like a rejected Black Mirror pitch: use AI to detect biological threats before they become pandemics. The company, founded by Kathleen McMahon (former head of Palantir's Life Sciences division) and Tess van Stekelenburg (Lux Capital partner with a computational neuroscience background from Oxford), secured backing from the OpenAI Startup Fund, Lux Capital, and Founders Fund.

The timing isn't subtle. International safety reports found that LLMs improved 80% in 2024 at providing instructions for weaponizing biological agents. Meanwhile, experts testified to Congress that we're maybe 2-3 years from AI dramatically widening the pool of actors who could engineer large-scale biological attacks. Apparently, the solution to AI-enabled bio-threats is more AI (because fighting fire with fire has such a stellar track record).

Valthos's approach involves analyzing biological data from wastewater monitoring and air surveillance systems, using machine learning to identify emerging threats, and designing medical countermeasures faster than traditional methods. They claim they can compress response times from months to hours. The nine-person team includes software engineers from Palantir, computational biologists from MIT's Broad Institute, and applied ML engineers from DeepMind, which is either reassuring or the exact cast you'd assemble for a techno-thriller.

Jason Kwon, OpenAI's Chief Strategy Officer, called this their first biosecurity investment, noting that "technology is moving fast" and the best way to keep up is "with more technology, more research, more startups." It's worth noting that current studies suggest today's AI models pose limited immediate risk for bio-weapon creation, but the gap between "limited risk" and "catastrophic risk" in AI has historically closed faster than regulatory frameworks can handle. So we're essentially building the immune system before the infection arrives, which is good planning unless we're accidentally teaching the infection new tricks in the process.

Researchers at McGill University just published a paper in the journal Device describing what they're calling the world's smallest 3D bioprinter, and the application is wonderfully specific: printing replacement tissue directly onto your vocal cords during throat surgery. The device, officially named MIISB (Minimally Invasive In Situ Bioprinter, because acronyms are mandatory in science), has a printhead just 2.7 millimeters wide, about the width of a sesame seed, and fits through the same laryngoscope surgeons already use for vocal cord procedures.

Swen Groen, the PhD student who led the project, designed it like a robotic elephant trunk (his words, not mine) controlled by tendon-like cables connected to a module outside the patient's body. The surgeon operates it in real-time, depositing a hyaluronic acid-based hydrogel that recreates the geometry of damaged vocal fold tissue. Professor Luc Mongeau, who leads the research group, admitted he initially thought "this would not be feasible" to make a flexible robot under 3 millimeters, which is academic speak for "I told my student this was impossible and they did it anyway."

The need is real. Between 3% and 9% of people develop voice disorders requiring surgical removal of cysts, polyps, or growths from their vocal cords, and the aftermath often involves fibrosis (scarring) that permanently damages voice quality. Current treatments involve injecting hydrogels through the skin of your neck, which is about as precise as it sounds when you're targeting a structure 1.5 centimeters long that vibrates thousands of times per second. The team tested their bioprinter on surgical training models and successfully reconstructed vocal fold geometry with repeatable precision.

Audrey Sedal, assistant professor of mechanical engineering at McGill, noted that "part of what makes this device so impressive is that it behaves predictably, even though it's essentially a garden hose." The next steps are animal testing followed by human clinical trials, funded by the U.S. National Institutes of Health. If approved, the technology could expand to other delicate internal tissues like the esophagus, colon, or heart. Basically, we're one step closer to surgery becoming less "remove the broken part" and more "print a new one while you're in there."

There are exactly two northern white rhinos left on Earth, both female, both living at Ol Pejeta Conservancy in Kenya, both functionally incapable of reproducing naturally. Their names are Najin and Fatu (mother and daughter), and they represent the endpoint of what happens when poaching and habitat loss meet a species with slow reproductive rates. But in September 2023, Colossal Biosciences partnered with an international consortium called BioRescue to attempt something between ambitious conservation and a biological Hail Mary: save the subspecies using IVF, stem cells, and genetic engineering.

The BioRescue team, led by Professor Thomas Hildebrandt at Germany's Leibniz Institute for Zoo and Wildlife Research, has been performing ovum pick-up procedures on Fatu since 2019. As of 2025, they've created 38 pure northern white rhino embryos using eggs from Fatu and frozen sperm from deceased males, all stored in liquid nitrogen. The first major milestone came in November 2023 when they achieved the world's first IVF rhino pregnancy in a southern white rhino surrogate named Curra. She carried the embryo for 70 days before dying from a bacterial infection, but the 6.4-centimeter male fetus proved the concept works.

Since then, three more embryo transfers have been attempted (July 2024, December 2024, May 2025), though none have resulted in ongoing pregnancies yet. The technical challenges are significant: they're transferring embryos into southern white rhino surrogates (a related but distinct subspecies), performing procedures that have never been done in rhinos before, and doing it all while Fatu ages past her reproductive prime. Scientists published the complete northern white rhino genome in PNAS in May 2025, which revealed crucial quality control issues in some stem cell lines and enabled the next phase: CRISPR gene editing to restore genetic diversity from museum specimens.

Colossal Biosciences, the company also working to de-extinct the woolly mammoth, is contributing genome sequencing, ancient DNA extraction, and gene editing expertise. Matt James, Colossal's Chief Animal Officer, suggested they're "3-4 years away" from producing a calf, though other scientists suggest decades may be more realistic. The ethical questions are substantial (should we invest millions in two individuals when habitat restoration could save thousands of other species?), but the technical achievement is undeniable. We're essentially building a reproductive system from scratch, using cells frozen years ago, surrogates from a different subspecies, and genetic material from museum bones. It's either conservation's most audacious rescue mission or an expensive demonstration that extinction has biological finality. We'll find out which one before 2030.

After 25 years without a major innovation in malaria treatment, Novartis and Medicines for Malaria Venture just announced results from a Phase III clinical trial that could represent the first new drug class since artemisinin-based therapies became standard of care. The drug, called KLU156 (ganaplacide-lumefantrine, or GanLum for short), achieved a 97.4% cure rate in the conservative regulatory analysis in a trial involving 1,668 participants across 34 sites in 12 African countries. More importantly, it works against drug-resistant malaria strains that are spreading across East Africa.

The KALUMA trial, presented November 12 at the American Society of Tropical Medicine and Hygiene meeting in Toronto, compares GanLum to Coartem (the current standard artemisinin-based treatment) in adults and children with uncomplicated malaria. The drug's mechanism is genuinely interesting. Ganaplacide disrupts the parasite's protein transport system inside red blood cells, killing the parasite through a completely different pathway than artemisinin. It's also active against gametocytes (the sexual stage that transmits to mosquitoes), potentially blocking transmission chains.

The timing is urgent. Artemisinin partial resistance has been confirmed in Rwanda, Uganda, Tanzania, and Eritrea, with some areas showing 40-50% of parasites carrying resistance mutations. We can’t forget the historical precedent. When chloroquine resistance spread across Africa in the 1990s, malaria deaths spiked to nearly 2 million annually. Professor Abdoulaye Djimdé, who coordinated the West African trial sites, called GanLum "the biggest advance in malaria treatment for decades."

But there's a significant problem: the drug tastes so catastrophically bad that approximately 20% of participants vomited compared to under 5% with the standard treatment (although this might not be just about taste). Treatment discontinuation rates were 10 times higher with GanLum. The formulation is a bitter powder delivered in sachets, and getting children to take it has proven challenging. Novartis says they're exploring taste-masking options and that improved administration instructions reduced the problem during the trial, but it's the kind of real-world implementation issue that can tank an otherwise excellent drug. The company plans regulatory submissions within the next 1-1.5 years, which means a new malaria treatment could reach patients by 2027, assuming they solve the vomiting problem. Because nothing says "breakthrough therapy" like "works great if you can keep it down."

While everyone's been obsessing over Ozempic and its relatives, Eli Lilly quietly published results in The Lancet for a weight-loss drug that works through a completely different mechanism and achieved 20.1% average weight loss at the highest dose. The drug, called eloralintide, is not a GLP-1 receptor agonist. It's an amylin receptor agonist, which means it mimics a completely different hormone pathway (amylin, which is co-secreted with insulin) to achieve appetite suppression and weight reduction. This matters because it represents an entirely separate biological mechanism, potentially offering options for people who don't respond well to semaglutide (Ozempic/Wegovy) or tirzepatide (Mounjaro/Zepbound).

The Phase 2 trial enrolled 263 participants with obesity or overweight (but not diabetes) across 46 U.S. research centers. After 48 weeks of once-weekly subcutaneous injections, the highest dose (9 mg) group lost an average of 20.1% of body weight, which translates to about 47 pounds for the average participant (baseline weight was 240 pounds). For context, Wegovy (semaglutide) typically achieves around 15% weight loss at standard doses, while Zepbound (tirzepatide) hits roughly 20-21% at its highest dose. So eloralintide is performing in the same efficacy neighborhood as the current best-in-class drugs, but through a pathway that's biochemically distinct.

The side effect profile was predictably gastrointestinal. Nausea occurred in 32.7% of all eloralintide participants (versus 13.5% on placebo), though the rate varied significantly by dose and escalation schedule. The 6 mg group had a rough time with 64.3% experiencing nausea, while the slower dose escalation (3-9 mg) group saw just 25%. About 10% discontinued due to adverse events overall. Secondary endpoints showed improvements in cholesterol, blood pressure, inflammation markers, and physical functioning scores.

Dr. Liana Billings, the lead author from Endeavor Health and the University of Chicago, emphasized that "obesity is a complex condition, and no single treatment works for everyone." The real value here isn't necessarily that eloralintide is better than existing options (it's comparable), but that it's different. When GLP-1 drugs don't work or cause intolerable side effects, having an alternative mechanism is clinically valuable. Eli Lilly plans to start Phase 3 enrollment by the end of 2025 and is also testing eloralintide in combination with tirzepatide, because apparently the next frontier in weight loss is just stacking different hormone pathways like biochemical Jenga. The drug won't hit the market for several years, but it's a reminder that the incretin pathway (GLP-1, GIP) isn't the only biological lever for appetite regulation. We're still figuring out how many switches the human body has for "stop eating," and it turns out there are more than we thought.

So, apparently, we're living in the timeline where AI companies fund biosecurity startups as a defensive move against AI-designed pandemics (very reassuring), surgeons will eventually be able to 3D-print replacement tissue mid-operation like some kind of biological autocorrect feature, and the concept of extinction is becoming more of a negotiable status than a biological endpoint. Meanwhile, malaria might finally have a new treatment option if we can figure out how to make it not taste like punishment, and the weight-loss drug market is expanding into new molecular territories because one receptor pathway is apparently never enough.

This is biotech in 2025: solving problems we created (AI bio-risks), problems we've had forever (malaria, obesity), and problems we thought were permanent (extinction), all while generating new questions about what we should be doing versus what we can do. The gap between those two questions has never been narrower, and we're moving fast enough that the ethics debates are happening in real-time alongside the clinical trials.

What does this mean for you? Probably that your vocal cords might one day be printable, that weight loss might come from an amylin agonist instead of a GLP-1, that the last members of a species might not actually be the last if we're determined enough, and that someone somewhere is already thinking about the next biological threat that doesn't exist yet. The future keeps arriving ahead of schedule, and biotech is still figuring out whether to hit the gas or the brakes.

If you found this useful (or at least entertaining while being accurate), forward it to someone who needs to know that science is simultaneously more advanced and more chaotic than they thought. And if someone forwarded this to you, subscribe because apparently this is how we keep up with biology now.

Keep questioning everything (especially the timeline),
Prateek & Jere