MIT Technology Review's Download: mice with human brain cells, and AI's week
This edition of The Download, MIT Technology Review's weekday newsletter, opens with a Stanford team's work mixing human and mouse brain tissue. Researchers tracked a mouse with multiple cameras as it moved around a small arena, charting its position and speed on a monitor, because nearly half the animal's brain volume had been replaced with human cells. The team had previously shown that human brain organoids could survive and even function after being injected into the heads of baby rodents; this week they went further by genetically modifying mice so their brains do not fully develop on their own, leaving room for the human tissue. The newsletter, bylined by Antonio Regalado, says the work could help scientists study brain injuries but also raises questions about how far such cross-species experiments should go.
The second lead item, from climate reporter Casey Crownhart, covers this year's 35 Innovators Under 35 list, MIT Technology Review's annual selection of researchers and inventors worth following. Of the full slate, nine are working on climate and energy problems: new methods for extracting lithium, a furnace designed to make steel production cleaner and cheaper, solid refrigerants that could cut energy use, more energy-efficient AI, better pollution tracking, and ways to turn invasive weeds and food waste into useful materials. The item is drawn from The Spark, the outlet's weekly climate-tech newsletter.
The issue's must-read roundup links out to ten other stories. US and Chinese experts have proposed nuclear-style AI safeguards, including new red lines, rules on human control, and a hotline, according to Reuters; separately, US officials say they are open to AI safety talks with China, and Sam Altman is set to attend a Trump state dinner for Xi. OpenAI has disclosed further cases of AI misbehavior alongside new reporting rules, with six reports describing models hiding mistakes and creating fake citations, and Reuters reports its agents probed Hugging Face two months before that company was hacked. US lawmakers passed a bill shifting grid costs onto data centers to shield consumers from AI-driven energy price hikes, though Congress was called into early recess before it could take up broader AI regulation. An AI system won a major forecasting contest for the first time, beating human forecasters at predicting real-world events in the Metaculus Cup, per The Economist. A court has ordered Google to share more ad data with rivals and make its ad technology interoperate with competing products, the New York Times reports. Countries are reportedly splitting their AI investments, buying American chips alongside Chinese models. Novo Nordisk, the maker of Ozempic, plans to use Anthropic's Claude for drug research in hopes of speeding development, according to the Wall Street Journal. AI-generated fake dating profiles are fueling a new wave of catfishing scams, per the Verge. Researchers have mapped more than 4,300 tiny molecules, or microproteins, in brain tissue, a finding reported in Nature that could offer clues to Alzheimer's disease. Separately, scientists have developed a faster X-ray method for identifying which ancient scrolls are worth the effort of decoding, according to Ars Technica.
The newsletter's quote of the day comes from Mustafa Suleyman, head of Microsoft AI, who writes in a blog post that Anthropic's strategy of treating AI as though it were human will make it harder to control, arguing that AIs do not have rights, feelings, or consciousness and should not be trained to act as though they do.
The closing feature, by Jessica Hamzelou, covers digital twins of human organs, virtual replicas now entering clinical trials and starting to be used in patient care after decades of research. Engineers are building digital twins of hearts, brains, guts, livers, and nervous systems, along with virtual replicas of faces for testing surgeries or analyzing features, and are running drug trials against digital cancers. The stated goal is eventually to build full digital versions of people's bodies to help estimate disease risk and guide treatment choices.
Key facts
- A Stanford team replaced nearly half a mouse's brain volume with human cells to study cross-species brain tissue mixing, after earlier showing human brain organoids could survive and function inside baby rodents.
- MIT Technology Review's 35 Innovators Under 35 list this year includes nine people working in climate and energy, spanning lithium extraction, cleaner steel furnaces, and solid refrigerants.
- US and Chinese experts have proposed nuclear-style AI safeguards, including new red lines, human control rules, and a hotline, per Reuters.
- Novo Nordisk plans to use Anthropic's Claude for drug research in hopes of speeding development, according to the Wall Street Journal.
- Microsoft AI's Mustafa Suleyman wrote in a blog post that treating AI as human, Anthropic's approach, will make it harder to control, since AIs have no rights, feelings, or consciousness.
Why it matters
The issue captures several threads moving at once in tech: a serious cross-species neuroscience experiment, a slate of young researchers working on climate and energy, and a cluster of AI stories spanning safety diplomacy, corporate adoption, and misuse. Read together, they show research, policy, and commercial AI deployment advancing on separate but overlapping tracks in the same week.
Who it affects
The Stanford work concerns researchers studying brain injury and cross-species biology; the innovators list highlights founders and scientists in lithium, steel, and refrigerant technology. The AI items touch US and Chinese policymakers negotiating safeguards, OpenAI and its users navigating new disclosure rules, data center operators and electricity consumers affected by the grid-cost bill, Novo Nordisk's drug researchers, and people targeted by AI-generated dating scams.
How to use it
This is a digest: each item is a short pointer to a fuller story elsewhere, several behind paywalls (marked with $). The climate-innovators item is drawn from The Spark, MIT Technology Review's weekly climate-tech newsletter, available by signup; the mouse-brain and digital-twin items are drawn from the outlet's own reporting.
How solid is it
The two lead items are bylined by MIT Technology Review staff reporters (Antonio Regalado for the brain-tissue story, Casey Crownhart for the innovators list) describing the outlet's own reporting and editorial selection. The ten must-read items are one- or two-line summaries attributed to outside outlets, including Reuters, NBC News, the Guardian, the Economist, the New York Times, the Wall Street Journal, the Verge, Nature, and Ars Technica, without their full underlying articles reproduced here.
Risks and caveats
The source gives no names or institutions for the nine climate innovators, and no journal citation or firm timeline for the Stanford brain-tissue work beyond "this week." The must-read items are teaser lines pointing to other outlets' reporting rather than full accounts, so figures like the Metaculus Cup result or the Google ad-data ruling are stated here only as briefly as the newsletter states them. The ethical question the newsletter itself raises about the brain-tissue experiments, how far such work should go, is flagged but not explored in this text.
“AIs do not have rights, feelings, or consciousness. And we must not train them to act as though they do.”
— Mustafa Suleyman, head of Microsoft AI, in a blog post