🧠 Sam Altman Wants Inside Your Head (No Surgery Required)

November 2025. The month when Sam Altman decided brain chips weren't invasive enough without the chip part, Japanese researchers made telepathy real (sort of), and we lost one of the most brilliant and problematic figures in modern biology. Also, turns out we might actually be close to defeating Huntington's disease, which would be nice positivity for a change.

NEWS

Sam Altman - fresh from his "definitely not a coup" OpenAI drama - is co-founding Merge Labs, a brain-computer interface startup seeking $250 million at an $850 million valuation. The twist? No Neuralink-style brain surgery required. Instead, they're combining ultrasound with gene therapy in an approach that sounds like science fiction had a baby with a NIH grant application.

Here's how it works: inject genes into your brain cells that make them respond to ultrasound waves, then use an external ultrasound device to read neural activity. It's sonogenetics, and it's being pioneered by Mikhail Shapiro, a Caltech chemical engineering professor whose research has already demonstrated 100 μm resolution brain imaging that could decode movement intentions before subjects moved.

Altman's vision? "I would like to be able to think something and have ChatGPT respond to it." Read-only access to your thoughts, basically an office inkjet printer, but make it telepathy. He explicitly stated he "would definitely not sow something to my brain" that kills neurons, taking a clear shot at his former OpenAI co-founder Elon Musk's Neuralink.

There’s a catch, though (there always is). This technology is obviously years from human trials and requires extensive gene therapy that hasn't been tested in human brains. Your private thoughts remain private for now. The company hasn't officially launched, and OpenAI hasn't committed to funding despite the plan for their VC arm to lead the round. Privacy experts are appropriately freaking out about "neural data access and the risk of hacking human thoughts", which is either forward-thinking caution or the setup for a Black Mirror episode.

RESEARCH & NEWS

A sealed plastic bag filled with synthetic amniotic fluid kept fetal lambs alive and developing normally for 28 days, and it's finally approaching human trials after years of regulatory wrangling. The technology - called EXTEND or "BioBag" - was developed by fetal surgeon Alan Flake's team at Children's Hospital of Philadelphia and commercialized by Vitara Biomedical, which just raised $50 million in Series B funding in November 2024.

The lambs used in the landmark 2017 study (biologically equivalent to 23-24 week human preemies) opened their eyes, grew wool, showed normal breathing movements, and developed healthy lungs and brains while floating in fluid. One lamb was bottle-fed and lived for over a year on a Pennsylvania farm. What’s the sauce? A pumpless circuit where the fetal heart drives blood flow through an artificial placenta for oxygen exchange, while lungs stay fluid-filled in their natural fetal state. No mechanical ventilation means no ventilator-induced lung injury, which causes most of the devastating complications in extremely premature babies.

Compare that to current NICU care: at 22 weeks of gestation, survival is approximately 10%. At 23 weeks, just under 56%. Survivors face chronic lung disease, brain hemorrhage, cerebral palsy, and developmental delays.

The FDA held a Pediatric Advisory Committee meeting in September 2023 to discuss the regulatory pathway, including thorny ethical questions: How do you get informed consent from desperate parents in crisis? How do you design trials when we can't predict which babies would survive with standard care? And perhaps most challenging, the technology requires C-section delivery directly into the bag, adding surgical risk for mothers. Human trials could begin in the coming years, though Vitara hasn't disclosed a specific timeline. Meanwhile, a separate Dutch startup called AquaWomb is developing its own version but remains years behind.

RESEARCH & NEWS

Professor Charles Xavier is real, he's Japanese, and he works at NTT Communication Science Laboratories. Tomoyasu Horikawa's "mind captioning" technology, published in Science Advances in November 2025, can watch your brain activity and write sentences describing what you're seeing or remembering. We're talking complete structured sentences, not just "person... dog... ball" word salad.

The system works in two stages: First, it uses fMRI brain scans to translate neural activity into semantic features while subjects watch 2,180 unique video clips. Then it uses iterative optimization with a masked language model to generate descriptions. Start with gibberish, run 100 iterations, and end up with coherent sentences. For a waterfall video, the system evolved from "spring flow" to "above rapid falling water fall" to "a person jumps over a deep water fall on a mountain ridge".

The accuracy? ‘Pretty good’ with ~50% success in identifying the correct video from 100 candidates when subjects actively watched videos. That's 50 times better than random chance. For recalled memories (eyes closed, just remembering), accuracy dropped to nearly 40% but was still far above chance. The system captures relational meaning - "dog chasing ball" versus "ball chasing dog" - not just vocabulary lists.

Before you mentally review all your inappropriate thoughts today, you can't read random people's minds. Each of the six subjects spent ~17 hours in an MRI scanner for personalized calibration. The technology only works on visual content subjects is actively being viewed or recalled, not private thoughts or abstract concepts. It's translating nonverbal visual representations, and it doesn't even need traditional language brain areas to work - Horikawa's team achieved nearly 50% accuracy with language regions excluded after all.

Potential applications include communication devices for stroke survivors with aphasia or ALS patients, where motor-based brain interfaces fail. The ethical implications are appropriately terrifying. Horikawa acknowledges risks of "unintentionally disclosing primitive thoughts before individuals have chosen to verbalize them" and calls for explicit informed consent requirements and mental privacy safeguards. Your secrets are safe for now, but the gap between science fiction and "please sign this consent form" is closing fast.

NEWS

James Watson died on November 6, 2025, at 97 in a Long Island hospice, closing the book on molecular biology's most brilliant and problematic figure. He co-discovered DNA's double helix structure at age 24, one of the most important scientific breakthroughs of the 20th century, and spent his final years as a pariah, stripped of honors by the institution he built.

The 1953 discovery at Cambridge's Cavendish Laboratory with Francis Crick launched the molecular biology revolution. Their famous Nature paper contained the understatement of the century: "This structure has novel features which are of considerable biological interest." It earned Watson, Crick, and Maurice Wilkins the 1962 Nobel Prize and enabled everything from genetic engineering to personalized medicine.

But the discovery was built on Rosalind Franklin's "Photograph 51" - the X-ray diffraction image that clearly showed DNA's helical structure. Wilkins showed it to Watson without Franklin's knowledge or permission in early 1953. Watson later wrote that seeing it made "my mouth fell open and my pulse begin to race." Franklin published her own paper in the same Nature issue but was positioned as corroboration, not co-discovery. She died of cancer in 1958 at 37, four years before the Nobel, which isn't awarded posthumously. Watson's 1968 memoir "The Double Helix" portrayed her in sexist terms, calling her "Rosy" (a nickname she never used) and critiquing her appearance.

Watson transformed Cold Spring Harbor Laboratory from a struggling facility into a world-class research powerhouse during his tenure as director (1968-1994). He launched the Human Genome Project and established ethical frameworks for genomic research. His textbook "Molecular Biology of the Gene" revolutionized science education.

Then came the fall. In 2007, Watson told the Sunday Times he was "inherently gloomy about the prospect of Africa" because "all our social policies are based on the fact that their intelligence is the same as ours - whereas all the testing says not really." Cold Spring Harbor suspended him. He apologized and resigned. In 2019, he told PBS his views had "not changed at all". Cold Spring Harbor revoked all his honorary titles and severed all ties, calling his remarks "reprehensible" and "unsupported by science."

As Venki Ramakrishnan, a Nobel laureate, noted in Watson's obituary: "Few people have influenced modern molecular biology and genetics as much as Jim Watson... so it is a pity that his later life was marred by his extreme views on race, gender, and genetic determinism. In the end, his contributions will long outlast his prejudices." DNA made him, and DNA unmade him.

RESEARCH

For the first time in history, a treatment has actually slowed Huntington's disease progression - not just masked symptoms, but demonstrably changed the disease course. uniQure's AMT-130 gene therapy met its primary endpoint with 75% slowing of disease progression (p=0.003) at 36 months in the high-dose group. One medically retired HD patient returned to work. This is the breakthrough the Huntington's community has waited decades for.

AMT-130 uses an AAV5 virus to deliver microRNA instructions directly into the brain's striatum via 8-10 hour stereotactic neurosurgery. The microRNA permanently silences the huntingtin gene, preventing cells from producing the toxic protein that destroys neurons. One injection, lifelong effect (in theory). The Phase I/II trial treated 29 patients total; the high-dose group of 12 patients with 36-month data showed significant slowing on composite clinical measures.

Beyond the primary endpoint, AMT-130 showed 60% slowing on functional capacity (p=0.033) - the ability to handle daily activities like finances and work. Cerebrospinal fluid biomarkers showed an 8.2% reduction in neurofilament light protein, a marker of neuronal damage that typically increases 20-30% over three years in untreated patients. Professor Ed Wild at UCL, a principal investigator, said his "patients in the trial are stable over time in a way I'm not used to seeing in Huntington's disease".

A big reality check hit in November 2025: FDA is no longer aligned with uniQure on using external control data as the basis for approval. The trial used propensity-matched natural history data from 154 patients in Enroll-HD as comparators rather than concurrent placebo controls. This design decision, made to avoid denying treatment to patients in a fatal disease trial, may delay approval by years. The sample size is small (12 patients), the data haven't been peer-reviewed yet, and manufacturing/surgical capacity would need a massive scale-up.

But make no mistake: this fundamentally changes what's possible for a disease that affects roughly 75,000 people in the US and Europe and has had zero disease-modifying treatments since its genetic cause was discovered in 1993. Huntington's is caused by CAG repeat expansion in the huntingtin gene - autosomal dominant, 50% inheritance risk, fatal within 15-20 years of symptom onset. Current treatments like tetrabenazine only reduce chorea (involuntary movements) without slowing the underlying neurodegeneration. Professor Ed Wild told reporters: "Today we get to move Huntington's disease into the column headed 'treatable'".

Some weeks, biotech news reads like a fever dream written by optimistic futurists and anxious ethicists locked in the same room. This was one of those weeks. We've got brain-reading startups, artificial wombs, actual mind-reading technology, the death of a giant, and a genuine breakthrough in neurodegenerative disease - all while navigating the ethical minefields and regulatory hurdles that make this field endlessly fascinating and occasionally maddening.

This newsletter reaches over 2,000 of you, and more than 70% are actively reading because, apparently, you people enjoy having your minds blown and your expectations complicated on a weekly basis. If someone in your life would appreciate learning that Japanese scientists can literally read thoughts, but your private internal monologue remains blessedly private (for now), forward this email. They'll either thank you or start wearing a tinfoil hat. Either way, you've added value to their week.

Keep questioning everything (especially the things that sound too good to be true),
Prateek & Jere

P.S. Consider that Sam Altman wants read-only access to your thoughts. Elon wants read-write. We're rooting for neither. Some thoughts are just meant to stay private.