2026-10-28 04:50:16
By chilling roundworms, scientists traced a molecular pathway hinting at one way cold might extend lifespan.
From ice baths and cold showers to whole-body cryotherapy, cold exposure has long been hyped as a wellness shortcut. The supposed benefits include boosted dopamine, weight loss, stronger immunity, or even slower aging.
Scientific evidence for these claims is scant, but a telltale link between temperature and longevity has intrigued scientists for over a century. Lowering ambient temperature by just a smidge can extend lifespan in the roundworm C. elegans, fruit flies, and even mice. Plenty of mechanisms could explain the curious effect, but scientists are still largely in the dark.
A new study may shed some light on the matter. When scientists grew roundworms at moderately cold temperatures, the worms steadily ramped up production of BH2, a helper agent for many enzymes, including those that build brain chemicals. Though initially produced in our muscles, BH2 roams to other parts of the body too, including the intestines and brain. Here, in the study, it boosted tyrosine production and contributed to a longer lifespan.
Mapping a longevity pathway in worms isn’t just about satisfying scientific curiosity. When the scientists supplemented human cells with BH2 or tyrosine, the treatments fended off toxic protein clumps associated with neurodegenerative diseases.
But don’t go ordering tyrosine just yet. A biological response in cells doesn’t automatically translate into a health benefit, especially one as complex as slowing aging. While worms are a useful model in longevity studies, they’re a far cry from humans. That said, an existing drug used to treat a metabolic condition can tweak the pathway, and examining clinical data from patients taking the drug could offer a useful test to see if the mechanism holds up in people.
More broadly, the findings point to muscles as active players in the aging process. Beyond movement, their chemical output could change how other tissues age. The findings “highlight the muscle’s ‘metabolome’—its full set of metabolites—as a potential source of signals that promote longevity,” wrote Hanseul Lee and Seung-Jae Lee at the Korea Advanced Institute of Science and Technology, who were not involved in the study.
As time ticks by, basic cell functions gradually falter. DNA’s protective end caps, called telomers, fray. Mutations accumulate in our genetic blueprint. Chronic inflammation flares across the body. Senescent “zombie” cells bathe healthy tissues in a toxic chemical brew. Cellular energy factories and waste disposal plants begin to sputter.
Dubbed the hallmarks of aging, these processes only partly depend on genetics. Environment, social factors, and lifestyle—diet, exercise, exposure to toxins—also shape our aging trajectory.
Temperature seems like an odd addition to that list. Yet nature is littered with examples. Hibernating mammals generally live longer than those who don’t. Yellow-bellied marmots, for example, spend up to seven months a year hibernating, lowering their body temperature and throttling metabolism. Scientists have found their biological aging essentially stalls during hibernation.
These observations inspired the rate-of-living theory, which suggests that the lower a creature’s metabolic rate, the longer it lives. Recent work has given the century-old idea a nuanced update by exploring how cooler temperatures directly activate mechanisms that counter some of the hallmarks of aging. But there are caveats.
Take mitochondria, our cells’ energy factories. With age, they become increasingly dysfunctional. A chill activates brown fat, which is packed with mitochondria, ramping up energy use and potentially overall metabolic health. In a small study of a dozen people exposed to mild cold for up to eight hours, most showed signs of better blood sugar regulation. But these were short-term effects, not evidence for lasting protection. Cold can also boost mitochondrial numbers, though whether they protect against aging in humans remains to be seen.
The immune system may also respond to a chill. A study in healthy people and those with inflammatory disease, found cold exposure slashed immune molecules called cytokines, which often fuel inflammation. A different study found lower temperatures slowed the accumulation of reactive oxygen species (little “bullets” that damage cellular components, including DNA).
A cellular waste removal plant, the proteasome, also seems to benefit from air conditioning. In 2023, David Vilchez at the University of Cologne and colleagues found that moderately cold temperatures boosted proteasome activity in both C. elegans and human cells. The newly invigorated cleanup crew chomped proteins known to aggregate into clumps involved in neurodegenerative diseases.
“Lowering body temperature is one of the most effective mechanisms to extend longevity,” the team wrote at the time. But translating those ideas into a practical strategy isn’t straightforward—continuously maintaining a low body temperature is a hard sell. Mimicking those effects in a pill, however, might be easier to swallow.
In the new study, Vilchez’s team compared metabolites in C. elegans grown at a comfortable 20 degrees Celsius (68 degrees Fahrenheit) with those in worms kept at temperatures five degrees lower. On average, the chilled worms lived roughly 15 percent longer.
The levels of one metabolite, BH2, skyrocketed in the cold. The molecule helps a host of enzymes do their jobs, and genetically blocking its production wiped out the extra longevity.
The team found BH2 was mostly produced in muscles and transported to other organs. In the intestines and brain, it shifted into another form that pushed cells to make more tyrosine, an amino acid often found in meats, fish, diary, and soy. Even without a cold snap, feeding tyrosine to worms unable to make BH2 helped them live longer.
“We were genuinely surprised to find the supplementation significantly extended lifespan, even at standard and warmer temperature,” said Vilchez in a press release.
The team then tested BH2 in human neurons. Mutant proteins related to Huntington’s disease, amyotrophic lateral sclerosis, and dementia often trigger cell death, but cells treated with BH2 largely escaped this fate. Meanwhile, they better maintained protein homeostasis, the delicate balance between making, folding, and clearing proteins that goes awry with age.
These findings suggest a way to mimic the benefits of cold without having to tolerate the shivers or dangerous health effects. But a designer pill is a long way off. Safety is one concern. In mammals, a balance between BH2 and its active version regulates reactive oxygen species. Supplementation may tip that balance, causing cells to pump out more “bullets” and ironically accelerate aging. And because the pathway controls other enzymes, unexpected side effects could do more harm than good.
For now, the takeaway isn’t to stock up on BH2 or tyrosine as a fountain of youth. Rather, the results are a good starting point for the development of metabolite-based treatments for neurodegenerative disorders. And instead of going after therapies that target the aging body as a whole, the results suggest focusing on one tissue could be enough.
“These findings suggest that specifically targeting a single tissue, such as muscle, could provide organism-wide protection. This opens up new avenues for therapeutic approaches,” said Vilchez.
The post Study Could Help Explain Why Colder Body Temperature Extends Lifespan in Many Species appeared first on SingularityHub.
2026-10-27 06:41:53
It’s widely believed AI’s inner workings are largely inscrutable. A former OpenAI researcher now says this narrative could be wrong.
The idea that the neural networks underpinning modern AI are “black boxes” almost impossible to understand has become received wisdom. But a provocative blog post from a researcher who played a key role in setting that narrative says these models may be far more interpretable than we thought.
The opacity of today’s leading AI models is becoming an increasingly pressing problem as the technology seeps into ever more areas of our lives. Recent incidents where agents from leading developers escaped their sandboxes to go on hacking sprees have only boosted the sense of urgency.
This has spurred interest in a research field known as mechanistic interpretability, which attempts to reverse engineer neural networks to learn what individual neurons and the circuits they create do. By detailing how these networks operate, researchers hope it’ll be easier to control and direct them.
Progress has been slow, however, and leading theories have suggested the way these networks learn leads to a tangled mess that is incredibly difficult to parse. Now though, Nick Cammarata, a researcher at AI startup Goodfire who helped lay the foundations of the field while at OpenAI, has published a 26,000-word blog post arguing that these theories are wrong and even the largest neural networks could be much simpler to read than expected.
“I think I have good news about the nature of neural networks,” Cammarata writes in the post. “Interpretability might be far more tractable, possibly in the near term, than we’ve assumed.”
The idea Cammarata is challenging is one he helped establish. In 2020, while studying a Google image-recognition model called InceptionV1, he and colleagues found neurons that seemed to respond to unrelated things—most famously one that fired on both cat faces and the fronts of cars.
The leading explanation was the concept of superposition, which suggested that when a network runs short of neurons, it starts lumping unrelated concepts together into the same neurons. And it specifically combines concepts that are so disconnected they’re unlikely to clash. This means that looking at an individual neuron tells you very little, because there’s no rhyme or reason behind what ideas it groups together.
But when Cammarata revisited the famous cat-car neuron, he realized both images shared a prominent V shape. Further tests showed the same neuron also responded to the corner of a pool table and the pointed nose of a Japanese bullet train.
Cammarata then went looking for similar explanations across the rest of InceptionV1 with the help of a specialized suite of tools and AI agents to help speed up the search. He concluded that unrelated concepts packed into neurons were usually more connected than they first appeared. This is often because a neuron starts out tracking one thing before gradually branching out into neighboring ideas as the network learns—a phenomenon he calls recursive generalization.
Cammarata has since applied the same approaches to modern language models. Preliminary results from the Qwen3-32B model suggest roughly 90 percent of its neurons track one main concept, potentially alongside a few closely related offshoots. Cammarata says he found similar results while analyzing the much larger, trillion-parameter Kimi K3 model, which he plans to detail in a forthcoming technical paper.
The blog post sparked huge debate in the AI community, including some rapturous praise from leading researchers at large AI companies. But not everyone is convinced. For a start, the blog post self-identifies as a working hypothesis rather than a rigorous empirical study. “I appreciate that he looked at thousands of examples, but even a large number here is not a substitute for rigorous evaluation,” Chris Potts at Stanford University told MIT Technology Review.
And others question how far understanding individual neurons can take you. In a series of posts on X, Jacob Hilton at the Alignment Research Center, pointed out that to truly understand what a neural network is doing you need to track and explain how they connect neurons into larger circuits.
Even Cammarata says he is surprised by the response the blog has received and admits some people have taken his findings too far. “It’s not infinitely good news; it’s not everything solved,” he told Tech Review. “There’s a bunch of open questions around, like how much the combinations of neurons matter versus just one at a time.”
Nonetheless, the work has reopened a central question in AI that many researchers considered settled. And while Cammarata’s forthcoming technical paper should give the field a clearer sense of how justified his optimism really is, there is now a sliver of hope that the intelligent machines we are building may be slightly less inscrutable than initially feared.
The post Is AI Really a Black Box? Neural Networks May Be Easier to Understand Than We Thought appeared first on SingularityHub.
2026-10-24 22:00:00
Hugging Face’s New AI Biology Model Lets You Chat With DNAEmily Mullin | Wired ($)
“The AI platform Hugging Face on Thursday released a new open model for understanding DNA that can run locally on a laptop. Developed in collaboration with researchers at Zhongguancun Academy in China and the University of Naples Federico II in Italy, the model is trained on genomic sequences and natural language so that people can ask questions about the effects of DNA changes on gene activity and health and get conversational answers in return.”
A First for CRISPR: Powerful Prime Editing Alters DNA Directly in the Human BodyHeidi Ledford | Nature
“The milestone treatment is part of a clinical trial of prime editing, a technique that can alter DNA with more precision than standard CRISPR–Cas9 methods can. In this case, prime editing was used to treat a condition called Wilson’s disease by rewriting DNA in liver cells, to erase a mutation that prevents them from eliminating excess copper.”
Welcome to the Valley of PainMatteo Wong | The Atlantic ($)
“This week could go down as the worst in the history of cybersecurity. …Yet what’s most alarming about the past few days is that by next year, this stretch of incidents—which some security experts are calling ‘Black October’—may be remembered as normal, or even tame. ‘It’s going to feel like, over and over and over again, we’re finding more and more bugs in very critical pieces of software,’ Matthew Prince, the CEO of the internet infrastructure and security company Cloudflare, told me.
Moderna and Merck Say Their mRNA Cancer Vaccine Cut Death, Recurrence by 42%Beth Mole | Ars Technica
“The results are a significant advance in melanoma treatment; Merck and Moderna are already engaging with regulatory authorities to get the vaccine approved. More broadly, the vaccine’s success bolsters hope that mRNA vaccines and individualized cancer-targeting medicines offer a new paradigm for cancer therapies.”
Startup Taps Underwater Volcano as Energy Source in ‘Proof Point’ TestJeremy Hsu | Ars Technica
“The world’s first demonstration of a subsea geothermal power plant has taken place near an active underwater volcano almost a mile below the ocean surface. It generated less than 10 seconds of net-positive electricity but showed the technical feasibility of tapping another form of renewable energy underwater.”
Is a Job Guarantee the Solution to Mass AI Unemployment?Nick Romeo | The Guardian
“In its most ambitious form, such a program would ensure that anyone who wants work is guaranteed a meaningful, decently paid job. By creating a public option for employment, its supporters argue, a job guarantee could protect millions of people from the financial and psychological damage of joblessness and mobilize a vast labor force to tackle urgent societal problems, from infrastructure and climate adaptation to elder care and childcare.”
This Touchscreen Is Made From Thread, and Scientists Want to Put It in Your ClothesVarun Mirchandani | Digital Trends
“Forget smartwatches. Scientists have developed a touchscreen that can be woven directly into clothing, letting users interact with digital content without carrying a rigid display. Researchers at the National University of Singapore (NUS) have integrated 54 specialized fibers into a sleeve to create what they describe as the world’s first touchscreen built into a garment.”
Cool, Cool, Cool, Palmer Luckey and Ben Lamm Made a Genetically Modified Military Dog StartupKate Knibbs | Wired ($)
“A new startup launched today with the definitely not terrifying mission of creating genetically enhanced dogs for the United States military, law enforcement, and emergency first responders. Warhound is the brainchild of Anduril’s Palmer Luckey and Colossal Biosciences cofounder Ben Lamm. The startup has raised $200 million to ‘rebuild America’s domestic supply of elite working dogs’ and reduce dependency on European dog breeders for military canines.”
‘Reverse-Engineer Anything’ Is the New Wild West of AIJason Koebler | 404 Media
“What we’re seeing is vibe coding, but instead of making something from scratch, people are using AI to take an existing piece of software and replicate its features, its functionality, or cloning it wholesale. In some cases, people are using AI to take a proprietary piece of hardware and install some version of open source software on it. All of this is some mix of ‘decompiling,’ ‘reverse-engineering,’ ‘hacking,’ vibe coding, AI slop, and stealing.”
Japan Launches Mission to Bring Home a Piece of the Martian Moon PhobosStephen Clark | Ars Technica
“The mission to the Martian moons builds on Japan’s success in leading the Hayabusa and Hayabusa 2 missions, the world’s first two spacecraft to bring back pristine specimens from asteroids. Scientists believe Phobos could itself be an asteroid that was captured into orbit around Mars billions of years ago. One of MMX’s science goals is finding out if that theory is correct.”
How a Legal Backlash Against Big Tech Is Forcing Big ChangesCecilia Kang | The New York Times ($)
“For decades, tech giants grew at breakneck speeds as Congress struggled to keep pace with the consequences of the companies’ rapidly booming businesses. But a new de facto regulatory regime has emerged over the past year as courts clobber the biggest internet companies. And for the first time, their product designs and advertising models—which critics say treat users as products—are costing the companies serious money.”
The post This Week’s Awesome Tech Stories From Around the Web (Through October 24) appeared first on SingularityHub.
2026-10-23 22:00:00
There is no big red button or glass box hiding a lever.
In the wake of high-profile hacking incidents involving advanced artificial intelligence agents, some politicians and business executives have been calling for an AI kill switch: a safeguard that could shut down these AI systems in an emergency.
The idea is that, if a model were to behave in dangerous ways, a kill switch would allow humans to regain control over that system by metaphorically flipping a switch from “on” to “off.”
Several efforts at AI regulation in 2026 have focused on the notion of a kill switch. These include proposed congressional legislation such as the AI Kill Switch Act, and California Gov. Gavin Newsom’s executive order aimed at reining in AI. President Donald Trump has opposed the idea of an AI kill switch as unnecessary regulation, and to date there has been no national regulation that includes a kill switch.
Experts such as Dylan Baker, a research engineer at the nonprofit Distributed AI Research Institute, have explained that one regulatory problem with the metaphor of the kill switch is that today’s advanced AI models involve a complex infrastructure of distributed technologies, which is beyond the control of a single kill switch.
Additionally, the metaphor of the kill switch also obscures the human decisions involved in the responsibility and oversight of an out-of-control model. The problem is that the metaphor does not account for the technological or human aspects of AI.
The metaphor of the kill switch is not necessarily new. In fact, it is rooted in the same rhetoric as the black box metaphor, which similarly blurs the complex aspects of AI by framing inner workings as confined and unintelligible. Referring to AI as a black box frames these systems in terms of restricted incomprehensibility: a monolithic stranger to human control and understanding.
At the University of Cambridge, digital humanities researcher Eryk Salvaggio argues that “the black box of AI cultivates useful confusion about what can and cannot be known” about the complexity of frontier models. The idea of the kill switch cultivates a similar confusion by obscuring the distributed infrastructure and human influence that are important aspects of AI. While these metaphors have opposite meanings—resisting control and regaining control—they invoke the same implicit argument about the nuts and bolts of AI.
Both the black box and kill switch are what cognitive linguist George Lakoff and philosopher Mark Johnson call structural metaphors. They rely on simple physical concepts—for example, object, substance, up-down, and in-out. Structural metaphors use one clearly delineated concept to help make another concept clearer. For example, black box uses the concept of an opaque container to give a clear structure to the concept of hidden or inscrutable inner workings.
Additionally, Lakoff and Johnson say that this type of metaphor can also hide some aspects of a concept by focusing attention on others. For example, the kill switch metaphor hides the complexity of advanced AI models by focusing on the fact that a model can be on or off.
What this means is that framing the operation of AI in terms of a black box and kill switch relies on the straightforward ideas of “box” and “switch” to conceal the complex operations of AI. Reining in AI becomes difficult on a policy level when discussions about AI regulation fail to account for the infrastructural and human aspects of advanced models.
There are two problems with the structural metaphor of an AI kill switch, both demonstrating that it misses the mark for a national safeguard. The first is that a kill switch makes the global infrastructure supporting AI appear centralized, when in reality it is a web of interconnected technologies spread across the globe.
The advanced models that companies like OpenAI and Anthropic are concerned about releasing rely on multiple computers and data centers that are spread out nationally. It is not as simple as pulling a plug, opening a box, or flipping a switch. The kill switch metaphor reduces the infrastructural entanglements of AI to outdated ideas about technology. For instance, in proposing his since-failed “AI Emergency Button Act” bill, Sen. John Kennedy compared the concept to kill switches in machines such as jet skis.
The second problem is that the kill switch idea keeps hidden the human decision-making involved in the training and deployment of frontier models. Rather than serving as a reminder that people are involved, the kill switch positions the human outside of the loop and ready to flip the switch during a catastrophic emergency. Human intervention is invoked as an afterthought, instead of a process that involves training, supervision, and approval.
For instance, accounts of how OpenAI’s model went “rogue” and hacked AI company Hugging Face depict the AI agents as though they operated on their own agenda. But experts have explained that not only did OpenAI personnel turn off safety mechanisms, but they also gave the model impossible tasks and left a door open to the internet. The role people played counters the “rogue” agent framing.
The kill switch metaphor is ill-suited to the challenge of regulating AI because there is no way to switch off an AI model without accounting for its distributed infrastructure and human influence. There is no big red button, emergency brake, or glass box hiding a lever. Effective national regulation of AI will not begin with any single switch or any one person’s finger.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
The post Why Calling for a ‘Kill Switch’ Is the Wrong Way to Think About Reining in AI appeared first on SingularityHub.
2026-10-22 22:00:00
A trio of papers showcases AI-designed molecular scissors and an upgraded tool that builds proteins atom by atom.
The protein universe is about to get bigger.
Scientists can now use AI to build proteins never before seen in nature. These designer proteins may become therapeutics, environmental sensors, or even the building blocks of synthetic life.
Despite enormous potential, however, AI tools still struggle to design complex proteins, like enzymes, from scratch. Unlike small proteins or peptide binders that disrupt cellular signaling, enzymes are shape shifters with a catalytic core. Getting the right shape is only half of the battle. The components must also quickly move, align, and interact at just the right moment.
But effective AI-designed enzymes may be here soon. This month, Nobel laureate David Baker’s lab at the University of Washington unveiled two types of AI-designed proteases. Proteases are like molecular scissors that cut other proteins at specific sites. The enzymes are notoriously tough to engineer, but designer versions could be made into powerful medications.
In test tubes, Baker’s synthetic scissors chopped up proteins associated with Alzheimer’s and ALS, a common motor disease. In cells, they acted like switches for inflammation-related proteins, activating them only after a precise snip and, in turn, changing how the cells behaved.
The findings open the door to destroying disease-related proteins long considered undruggable. “Proteases are just the beginning,” said study author Anqi Chen in a press release.
There’s another reason for optimism too.
In a separate paper, the Baker lab introduced RFdiffusion3, the lab’s latest generative AI model for protein design. Rather than mainly working at the level of amino acids, the tool zooms in on individual atoms, giving scientists finer control over their synthetic creations.
“The field is maturing from predicting and accurately building protein shapes and interactions to designing machinery that physically changes matter at the atomic level, in targeted and programmable ways,” said Baker in the press release.
Proteins are finicky things. Strings of molecules called amino acids twist into helices and sheets, then fold into intricate 3D shapes.
Not all amino acids are built the same. Some are relatively sleek; others sport sprawling branches called side chains. These branches influence how proteins fold and interact with their surroundings, affecting how drugs rev up or dampen their activity or change DNA and RNA behavior. Add in a protein’s ability to change shape and engineering one to behave predicably seems like an impossible dream.
With generative AI, the dream is becoming reality. Algorithms are expanding the protein universe, allowing scientists to engineer smaller or wholly new gene-editing tools, for example, or tiny antibodies that fight viruses or serve as antidotes to snake venom.
Enzymes, however, are tougher to crack. Many enzymes orchestrate an intricate dance between metals, target molecules, and other ingredients—such as water—to make a reaction happen. Any hiccups can torpedo the whole process.
Take proteases. These molecular scissors sever the chemical connections linking amino acids. Their core does the cutting, while other parts orient them so they dock to the target or otherwise control activity.
In two new studies, the Baker lab team designed two proteases. One, a zinc protease, uses a metal to drive the reaction. The other, a cysteine protease, forms a vulnerable chemical intermediate with the target, making the bond easier to break. Both strategies depend on getting specific amino acids into precise positions.
The team took inspiration from existing enzymes found in a crayfish’s digestive system and in yeast. With RFdiffusion2, their in-house design tool, the researchers built new protein backbones around each enzyme’s catalytic core to help it dock to the target. They used AI tools to select the best amino acids and built the protein around those structural constraints.
After repeating the process dozens of times, the team turned to AlphaFold3 to predict the resulting structures. Most looked nothing like known proteins, suggesting the AI wasn’t simply borrowing nature’s existing blueprints.
Thirteen of the synthetic cysteine proteases worked in a test tube test, a success rate of nineteen percent. But all were less active than the original yeast enzyme. The zinc proteases were even trickier. The team had to tweak amino acids near the catalytic core to accommodate another molecular player in the reaction. They found 36 percent of the 135 candidates could precisely cut their targets. One of these targets was amyloid-beta, a protein that forms toxic clumps in Alzheimer’s disease. The best designs made the reaction roughly 100 million times faster than when the protein target naturally broke down.
Designer proteases have a leg up compared to other protein-based medicines. Antibodies, for example, require a high dose because they only latch onto one target at a time. But with proteases, “you just need a small amount because each scissor molecule can cut many, many target molecules,” Baker said in a video provided to journalists.
But the overall low efficacy of the enzymes the team designed exposed an uncomfortable truth: Current AI tools still struggle to create more complex molecular machines.
To build better proteins, the lab developed RFdiffusion3, a protein design tool that zooms into the atomic world. Protein modeling algorithms, like AlphaFold3, already operate at this level, and some design tools can tweak atoms around catalytic cores. But most still focus on shuffling amino acids.
RFdiffusion3 uses atoms as the basic unit when generating the protein backbone and side chains, better accounting for the precise interactions that make enzymes work.
That’s not to say the model fully maps every atom. Each amino acid is represented by 14 atom slots—four for the backbone and 10 for side chains—that capture its structure and dynamics. Think of dots on a motion-capture suit. The dots don’t track every muscle twitch, but they do provide enough information to reconstruct movement.
This extra detail benefits enzyme design. Scientists can specify atomic positions around the catalytic core to better control the enzyme’s shape and chemical properties. The resulting blueprint more precisely maps out which amino acids to use and where to put them.
The team tested the tool on two types of newly designed enzymes in test tubes. One broke down paraoxon, an environmental pollutant. Using quantum chemistry data to model how the enzyme changes shape, the AI produced four active candidates on the first try, which the researchers further improved for speed and precision.
They also tested a zinc protease designed to cut a protein that, when misfolded and clumped, leads to kidney, heart, and nerve disease. Of 96 designs, 56 cleaved more than 10 percent of their target, a low but promising hit rate for enzymes designed from scratch.
Proteins don’t work in isolation. RFdiffusion3 can also design them in the context of other molecules—DNA, RNA, and drugs—accounting for the atomic details of their interactions. This could help researchers build final structures better tailored to their molecular partners.
As with all AI-designed proteins, each candidate will have to be vetted in cells. But the team is hopeful the upgraded tool could unlock more creative possibilities. “We anticipate RFD3 [RFdiffusion3] will be broadly useful for a wide range of biomolecule design challenges and applications,” they wrote.
The post This AI Model Designs Enzymes With Atomic Precision appeared first on SingularityHub.
2026-10-21 05:05:34
The treatment delivers a light-sensitive algae protein to the retina. Paired with special goggles, it can restore limited vision and lasts for years.
The patients sit perfectly still as a needle inches toward one of their eyes. After the injection, they’ll no longer be completely human. The syringe is filled with a gene therapy that installs a light-sensitive algae protein into failing retinas and could reboot some lost vision.
When scientists first proposed using algae proteins to restore human eyesight, the idea seemed like science fiction. Could alien proteins integrate into our intricate visual system? Even if they did, could the brain translate the resulting signals into sight, rather than nonsensical bursts of light?
A new study suggests yes, with caveats. The team treated people with retinitis pigmentosa, a genetic disease that gradually destroys light-sensing cells in the retina. Seven of ten patients showed improved light sensitivity. After patients donned goggles that deliver pulses of light to activate the algae proteins, the team tested their eyesight. Roughly half could reach for and touch objects or identify the orientation of a black bar, a common test of visual recovery.
The treatment is still far from restoring full vision. What patients saw was too blurry and the field of view too narrow to discern texture or fine detail. The boundaries of objects looked squiggly. But those who responded to the treatment continued to improve over roughly five years. One participant, whose story made headlines in 2021, can now recognize and count different objects, and white stripes on a crosswalk pop out.
The study, called PIONEER, is an ambitious clinical trial combining gene therapy with optogenetics. The first goal was to assess safety, and it met the mark. Most participants experienced a brief bout of inflammation in the eye receiving the injection. The only patient to suffer a severe side effect was quickly treated with medication.
The findings suggest the initial success wasn’t a fluke. The study “represents an important milestone for optogenetic vision restoration,” said study author Stefan Futterknecht at the Institute of Molecular and Clinical Ophthalmology Basel, in a press release.
Around two decades ago, Karl Deisseroth was hunting for new ways to control brain cells. Electrical bursts and chemicals, which are how neurons communicate, seemed like obvious choices. But nature had something stranger to offer: algae.
Algae produce a protein called channelrhodopsin. When activated by light, the protein changes neuron activity using a mechanism that works similarly in mammalian cells. In a foundational paper, Deisseroth’s team inserted the protein into rodent neurons, and voilá—the neurons could now be controlled using light.
The technology, called optogenetics, revolutionized neuroscience.
By genetically tagging specific neurons with channelrhodopsin, scientists can now dial the cells’ activity up or down like a dimmer switch. Optogenetic experiments have revealed brain circuits that shape feelings, memory, and cognition with unprecedented precision.
Researchers panning for more genes in different algae strains struck gold again. One channelrhodopsin variant they found, ChrimsonR, was particularly exciting. Unlike the classic version that’s activated by blue light, ChrimsonR responds to amber light, which penetrates tissues better and is less damaging. More recently, scientists have begun to use generative AI to redesign natural proteins to make them even more sensitive to light.
“Optogenetics provides opportunities for mapping the brain in a way that we could once only dream of,” said Per Svenningsson, Chair of the Nobel Committee for Physiology or Medicine, when the technology won the coveted prize this year.
Still, it has mainly been considered a research tool.
Retinitis pigmentosa gradually robs people of their eyesight. Researchers have linked over 100 genetic variations to the condition, which kills off light-sensitive cells in the retina.
“The replacement (or editing) of a single gene would not help most persons,” wrote Deisseroth in the New England Journal of Medicine. Instead, adding channelrhodopsin could offer a more universal solution by rewiring surviving neurons.
Early successes bolstered the idea. In one study, scientists equipped cones—a type of retinal light-sensing cell—with channelrhodopsin in mice with retinitis pigmentosa. The therapy boosted their light sensitivity. It also reactivated cones that had stopped responding to light in human retina samples.
The new study took a different approach. In advanced disease, cones may have already degenerated beyond the point at which they could be saved. So instead, the gene therapy targeted ganglion cells. These cells process signals from light-sensitive cells and pass that information to the brain, but don’t detect light themselves. Ganglion cells are, however, more resilient than cones. Add channelrhodopsin, and they could become substitute light sensors.
The team packed genetic instructions for ChrimsonR into a harmless viral delivery system and injected it into the poorer seeing eye of 10 patients aged 40 to 73. All were legally blind, and each received one of three escalating doses.
As the patients gradually built up levels of the algae protein in their eyes, the team faced another challenge: Translating visual scenes from the outside world into amber light that the engineered ganglion cells could recognize.
They developed specialized goggles that look a bit like Apple’s Vision Pro. The goggles use a camera to scan the field of view thousands of times per second, monitoring for light changes in every pixel. When the device detects a change, it pulses amber light from that pixel into the eye. Because our eyes naturally dart around—often without us noticing—this dynamic method may better mimic normal vision.
With the goggles on, 6 of 10 participants showed clinically meaningful improvement in light sensitivity, with more than a fourfold boost over baseline readings. One participant improved over 62-fold. Of the eight who agreed to further eyesight tests, roughly half could touch a notebook, grab a small staple box, and detect changes in a rotating bar.
Of those who initially responded to the treatment, improvements lasted for years. Their brains also seemed to adapt to the alien light sensors. In one test, participants were asked to open or close their eyes while looking at a tumbler. Electrical brain signals related to vision were more coherent with eyes open, suggesting their perception depended on improved visual input.
Any gene therapy carries risk. In an already worsening eye, inflammation and other side effects could cause further damage. Overall, the injection appeared relatively safe. Most participants experienced a temporary increase in eye pressure, but no detectable side effects throughout the body. The retina’s overall structure remained unchanged. One person suffered a blockage in a major eye artery that cleared within minutes after treatment with an FDA-approved eye drop.
For now, the participants still can’t read or see faces. The therapy reaches only a fraction of their ganglion cells, making sharper vision harder to restore. More advanced versions of channelrhodopsin with greater light sensitivity may increase resolution.
As for the hardware, the team is working on versions of the goggles that deliver more accurate and timely stimulation. Potential upgrades include an eye tracker, software to preprocess visual scenes, and even holographs.
“Overall, although optogenetic methods have been developed primarily for basic science discovery,” wrote Deisseroth, they’re now heading towards “diverse therapeutic approaches.”
The post Blind Patients Recover Partial Vision in Gene Therapy Trial appeared first on SingularityHub.
