Committee

MIT Technology Review Explains: Let our writers untangle the complex, messy world of science and technology to help you understand what’s coming next. You can read more from the series here. Late last month, Anthropic released a report arguing that the model GLM-5.3 poses a major cyberattack risk. The report’s core concern is that the powerful new model, released in August by the Chinese company Z.ai, is freely available for anyone to download and modify. Anthropic’s assertion is only the latest move in a long-running debate about whether frontier AI models should be made open, like GLM-5.3, or should stay locked behind the doors of the frontier labs.  These days, much of the debate about how to handle open models is playing out in public—and there’s a lot of money resting on its outcome. Nvidia CEO Jensen Huang, who has a lot to gain if people continue buying their own hardware to run open models, advocates enthusiastically for the United States to develop more of them, whereas Anthropic’s Dario Amodei, who runs a company that could be undermined by cheap, powerful models, advocates for a more cautious approach. But beneath all these commercial incentives, there is real substance to disagreements about open models. “I think a lot of the way you come down on the openness debate is influenced by how you come down on the AI debate,” says David Gray Widder, an assistant professor at the University of Texas at Austin School of Information. Those who are concerned about AI safety warn that open models can easily be misused by bad actors, whereas those who focus more on AI’s practical potential urge that this power should not sit behind the closed doors of just a few big companies, and that open models could help spread it more widely. There are a range of ways to handle open models between banning them entirely and releasing them willy-nilly. But any approach that a government or a society might take still involves trade-offs between safety and democratization. Deciding whether to support or limit the production of open models requires understanding those risks and benefits in depth. What are the benefits of open models? Let’s start with the basics. A closed model, like the models that power ChatGPT and Claude, can only be accessed through a website, app, or API. An open model, on the other hand, can be directly downloaded: The details about its internal structure, encoded in a long series of numbers known as the model weights, are publicly available. Such models are sometimes called “open-weight models,” to distinguish them from models for which even more information, such as details about the training procedure, is available.  Right now, closed models are more capable than open models—as of publication, the top eight models on the Artificial Analysis LLM leaderboard are all closed, though open models are only a handful of months behind in development. But there are several reasons why someone might opt to use an open model rather than a closed one. Because open models are downloadable, people can run them on their own computers or hard drives. They can even retrain them to behave differently, if they have the expertise and resources to do so.  If you’re an academic researcher working to understand and improve AI systems, you almost certainly need to work with open models in your research. When working with closed models, researchers can’t do much more than send prompts and observe the responses. But researchers can observe the inner workings of open models, as well as the “chain of thought” that the models use to think through a problem before answering. They can also combine multiple open models to build new systems, compress them to make them more efficient, and fine-tune them for new applications. Businesses, too, have reasons to prefer open models. Even if you depend on an external provider to run them for you, open models still typically cost less than closed models, according to a recent report from Mozilla. And if businesses want to use AI to process sensitive data or proprietary information, running a local model on their own hardware may be the best way to keep their operations secure. But perhaps the most important quality of open models is that they challenge the tight control the frontier labs have over how most of us use AI. “The clearest, biggest argument [for open models] is around preventing mass concentration of power—it is about making sure that this foundational technology is a public good,” says Elizabeth Seger, senior policy advisor at the Tony Blair Institute for Global Change. “I think we’re at this juncture where the question of who controls the infrastructure that allows AI to operate is as big of a question as who gets to control information on the internet.” Seger sees the future of AI playing out in different ways, depending on the role that open models have. In one possible future, frontier AI systems are controlled entirely by select people, companies, or governments, and as those systems grow more powerful, those groups have increasing control over what our world looks like. The world’s most sophisticated hacking models, for example, are currently all made in the United States, so it’s entirely up to US companies and the US government whether other countries can use them to shore up their own cyber defenses. If those companies remain in the lead, we could in principle wind up in a world in which the United States is capable of hacking any other country, but no one is capable of hacking it. Frontier open models could change that balance. “If every individual had a loyal-to-them AI system that was safe [and secure], I think that would be a much less concerning world than one in which they have to just rely on a government that says, ‘Yeah, we’ve got a super intelligent AI system, it’s got this plan, it’s inscrutable to all of you, just trust us,’” says Hamish Hobbs, director of AI policy

arXiv:2606.22807v4 Announce Type: replace Abstract: As retrieval systems scale, effective and efficient reranking becomes increasingly important. However, most existing encoder- and decoder-based rerankers jointly process every query–passage pair, tightly coupling their online computation and limiting deployment efficiency and flexibility. We present Lychee-Reranker, a fast but not late-interaction (FBNL) reranker that decouples query and passage computation while retaining expressive relevance modeling. Built on an encoder–decoder architecture, Lychee-Reranker pre-encodes passages using Matryoshka embedding pooling, while its decoder models system and user instructions together with query intent; cross-attention then captures fine-grained relevance between the resulting query context and passage representations. Together, these designs offer four key advantages: (i) efficiency from offline passage encoding, (ii) expressiveness from cross-attention, (iii) compactness from Matryoshka embedding pooling, and (iv) test-time compute through an adjustable compute budget. We instantiate Lychee-Reranker in three sizes, Nano, Small, and Large, with 0.27B, 1B, and 4B activated parameters, respectively. Extensive experiments on BEIR, MIRACL, and LMEB demonstrate strong reranking performance with superior efficiency. On BEIR and MIRACL, our models achieve competitive performance in multi-domain and multilingual reranking, on par with strong industrial rerankers such as the Qwen3/BGE-Reranker series. On LMEB-Dialogue, a compact embedding model paired with our Nano reranker, which has only 0.27B activated parameters, remains competitive with 7–12B embedding models. Lychee-Reranker was previously known as KaLM-Reranker-V1. Data and models are available at https://huggingface.co/collections/KaLM-Embedding/lychee-kalm-reranker-and-jev.

In a research station in central Panama, you can walk through a century of forest time, watching scrubby plots of adolescent trees give way to cathedral-­like canopies of mature growth. Forest ecologist Daisy Dent, a senior scientist at ETH Zurich, oversees this sequence of plots across the Barro Colorado Nature Monument, some of which date as far back as the 19th century. Not one of the trees she surveys was planted by human hands. She maps, measures, and tracks each trunk that emerges, revealing how waves of tree life rise from bare earth.  Tropical forests like these are our planet’s biggest terrestrial carbon sink, accounting for around 40% of carbon stored on land. As part of an international network of researchers, Dent is deciphering the networks of animal and plant interactions that enable some forests to recover after destruction. The goal is a low-cost, high-impact alternative to typical carbon-­crediting schemes, which plant huge areas with a single species of tree. For over a decade, her research has built a body of evidence that such large “monocrop” programs could be expensive mistakes—and that more resilient forests can rebound naturally in just a few decades, re-creating ecosystems that suck up vast quantities of carbon dioxide.  Few places are better to demonstrate this than Barro Colorado, a reserve covering the forested peninsulas and islands that formed when construction of the Panama Canal flooded the area in 1914. Under the management of the Smithsonian Tropical Research Institute, generations of ecologists have amassed more than a century of records.  That means Dent can trace why some plots rebound after clear-cutting for farming and forestry while others falter. Fruit-eating animals—including spider monkeys, bats, and parrots—are responsible for planting more than 70% of trees in tropical forests, and Dent’s dataset can tell which built each patch of trees by depositing seeds in their nutrient-rich droppings. Many of these creatures worked together to plant a diverse 70-year-old patch whose canopy is approaching the stunning structure and richness of an ancient jungle. In contrast, a patch of similar age populated by narrow trunks of a single long-lived species, Gustavia superba, hints that tropical rodents called agoutis are more or less the sole animal architects behind this more homogeneous forest.  The data she collects helps write a playbook for when to let ecosystems regenerate and when to give them a helping hand. One 2024 estimate in Nature found that some 215 million hectares of forest worldwide (an area larger than Mexico) could regenerate naturally. “Tropical forests are underpinned by this massive network of interactions,” Dent says—interactions between life-forms like the fungi that help trees take up nutrients, as well as pollinators and seed-distributors. Supporting those relationships is key to growing forests that lock up carbon and help sustain local communities and wildlife.  Filling in the gaps Just over a decade ago, as the landmark Paris Agreement was being codified in 2015, forest recovery seized global attention as a solution to the climate crisis. Global leaders rallied behind plans to regrow a trillion trees, after the ecologist Thomas Crowther highlighted their potential to capture a decade’s worth of global carbon emissions. But the biggest question has lingered: How, exactly, do you regrow two trees for every star in the Milky Way? While startups have proposed seed-spraying drones and nutrient-coated seed bombs, Dent has spent two decades studying what happens when we let forests rebuild themselves. As a PhD student around the turn of the millennium, she researched old-growth forests in northern Borneo but felt hopeless watching entire landscapes felled for palm oil plantations. Rather than despair, she turned to restoration. Then, as a postdoc in 2005, she joined Panama’s Prorena, a joint initiative between the Smithsonian Tropical Research Institute and the Yale School of Forestry and Environmental Studies, which aimed at determining how best to establish diverse native forests on clear-cut land. Back then, secondary forests—ones that grow naturally after the original forest was lost—had been “a bit of a niche research area,” Dent says. But as farming and human infrastructure have driven deforestation, resurgent forests now account for some 70% of all tropical canopies. Prorena initially followed the usual playbook of transplanting nursery trees to rangelands, but Dent began to question whether planting was necessary. In 2009, she published an encouraging paper showing that animal populations rebound quickly as forests naturally regrow. In 2010, she began working in Barro Colorado on the plots established by the pioneering American ecologist Julie Denslow. In 1994, Denslow identified plots that had been regenerating naturally for 20 to 100 years, and Dent would add younger ones to continue the sequence.  The site provides not just general data on forest regrowth but in-depth demographic information over time, says Robin Chazdon, a professor emeritus of ecology and evolutionary biology at the University of Connecticut who collaborated with Denslow when she established the plots. “It tells you more about the actual processes that are happening in the site,” she says.  For example, previous research had shown that fast-growing “pioneer trees,” the first to thrive in the damaged land, are eventually overtaken by larger carbon-storing trees. Dent’s research helped fill in gaps about the mechanics of how this happens—or doesn’t happen—in individual cases. In the Gustavia-dominated patch, for instance, stubborn grasses planted by cattle ranchers smothered most other trees before they could take root. These trees can hang around, crowding out the light, for 150 years, blocking the second wave of large-trunked trees.  To plant or not to plant Dent found herself well placed for the massive influx of interest in reforestation that began in 2015. She’d already become a principal investigator for 2ndFor, an international research network aiming to learn how best to manage secondary forests. Using data gathered from the Barro Colorado plots, the team authored a 2016 paper in Nature showing that a hectare of naturally regenerated secondary forest can capture 11 times as much carbon each year as an old-growth one, with young trees absorbing it quickly as they compete to grow toward the

This week my colleague Antonio Regalado had an interesting update on GLP-1 weight-loss drugs. According to research presented at an aging meeting in Boston, these drugs seem to affect at least some measures of biological age. Overweight and diabetic people who take GLP-1s have a biological age around two to three years younger than similar individuals who take a placebo, according to findings from Eli Lilly and Novo Nordisk, two pharmaceutical companies that make the drugs. I’ve heard people in the aging field describe GLP-1s as longevity drugs for a while now (even if they only benefit people who are overweight or diabetic in the first place). This latest announcement comes after we’ve seen a raft of other benefits associated with taking these drugs. But there are also some weird side effects. So far, GLP-1 agonist drugs—which repress hunger—have been approved for treating diabetes, chronic kidney disease, and obesity, and they can also be prescribed for some people who are overweight or at risk of certain cardiovascular diseases. But there is research to suggest they might also help people with substance-use disorders, and they’ve been investigated for potential benefits in neurodegenerative disease and cancer. And a lot of people are taking them. According to a recent Gallup poll, 11% of Americans are taking a GLP-1 drug for weight loss, up from 3% in 2024. And 15% of Americans say they’ve used one for weight loss at some point. They can be incredibly effective. But, like any drug, they can come with side effects. And we’re still figuring out exactly what those might be, and how widespread they are. The most common side effects are gastrointestinal—nausea, vomiting, diarrhea, and constipation. The drugs have also been linked to pancreatitis and gallstone disease.  Some claim that, in rare cases, the drugs might cause damage to nerves behind the eye. Dozens of people are suing Novo and Eli Lilly after developing a condition that can lead to sudden vision loss. They’re claiming that their symptoms were brought on by the GLP-1 drugs that these companies manufacture. Both companies dispute the claims. Research into the link is mixed. We’re still getting to grips with the effects of weight-loss drugs on hair and nails, too. When Savanna Vidal at the George Washington University School of Medicine and Health Sciences in Washington, DC, and colleagues assessed electronic health data from 67 US health-care organizations, they found that GLP-1 use was associated with an increased risk of hair loss. It’s still not thought to be common, however. And at a conference in Austria last week, another group presented the results of a smaller study suggesting that 66% of GLP-1 users reported at least one nail disorder—in some cases involving nail detachment. Nail conditions were also fairly common among people who didn’t take the drugs, though. The authors don’t claim that the drugs are causing nails to fall out (as I’m sure 11% of the American population will be relieved to hear). The side effects might look different in children, but there’s not much data on those at the moment. Still, pediatric side effects will be something to watch given that, between 2019 and 2026, there was a 310-fold increase in GLP-1 prescriptions for eight- to 11-year-old children with obesity in the US. Child prescriptions are still rare, but they’ve been rising. On Wednesday, the World Health Organization released its first guidelines on managing obesity in children and adolescents. Despite the scale of the issue—the WHO estimates that 70 million children aged five to nine were living with obesity in 2024—the organization explicitly says it does not recommend pharmacological treatment for kids under 10. “This recommendation accounts for the lack of evidence on both benefits and the potential adverse events and consequences of the different medications and/or their active ingredients on normal growth, development, physiological parameters, and mental health,” the guidelines state. I get press releases about GLP-1s on an almost daily basis. There’s been a ton of research into these drugs, and there are probably hundreds of other studies underway. There will likely be plenty more on this from The Checkup as the evidence evolves. This article first appeared in The Checkup, MIT Technology Review’s weekly biotech newsletter. To receive it in your inbox every Thursday, and read articles like this first, sign up here.

The moment Trevor Wischnewsky heard that drones were delivering pizza and sushi around his Texas neighborhood, his mind was made up. “It was super fascinating,” he says. “I just immediately wanted to be a part of it.” Wischnewsky is now an RPIC, or “remote pilot in command,” at a company called FlyTrex. An RPIC is like an air traffic controller. But instead of airplanes, this new class of aviators tend to fleets of hexa­copter drones that ferry cargo to homes and businesses within a three-mile radius. Not only does he ensure the safe passage of FlyTrex’s parcels, but Wischnewsky coordinates traffic with a growing swarm of delivery drones dotting the skies over Dallas from competitors like Zipline, Amazon, and Alphabet’s Wing. Here’s what it takes to land the gig. Aviation experience When he applied to FlyTrex, Wischnewsky was in training to become a commercial pilot and flight instructor, and he was already certified by the Federal Aviation Administration as a drone pilot. The company requires this credential, along with a weeks-long FAA-approved training course, to ensure that a pilot is prepared to take over a drone if need be. Though he doesn’t pilot the craft, Wischnewsky monitors the delivery process from FlyTrex’s operations center with the help of inputs from GPS, lasers, and a range of sensors. A collaborative mindset At first, the biggest challenge was trusting the technology. “On the manned-­aircraft side, they tell you ‘Don’t trust the automation,’” Wischnewsky says. The Dallas–Fort Worth area, though, is a testing ground for unmanned aircraft systems traffic management (UTM), a network that allows autonomous drones from FlyTrex and its competitors to share their coordinates in real time to help avoid collisions even beyond the pilot’s line of sight. Once Wischnewsky makes sure a drone is flight-ready, its path is unimpeded, and the weather conditions are clear, UTM takes care of the rest. Ability to stay calm under pressure At his busiest, Wischnewsky will have up to 10 drones on his watch at the same time. “You’re there as a safety net,” he says. Even with automation, keen situational awareness is required, and the most common hazard is the weather. Wischnewsky recalls a particularly stressful day: A drone had been in the air for only seconds when the wind picked up so suddenly that he had to intervene. He returned it safely, preventing an errant ice cream cake from hitting the sidewalk. Eamon Whalen is a reporter and researcher based in San Francisco.