Welcome to Peer Review'd, the podcast where we dig into the latest science news and break it down for curious minds everywhere. I'm your host, and today we have a packed episode full of discoveries that range from the microscopic to the oceanic, from ancient life to cutting-edge AI. Let's dive in.

We're going to start with some exciting news in medicine. A groundbreaking clinical trial has found that metformin, the affordable, widely-used medication typically prescribed for type 2 diabetes, may actually benefit people living with type 1 diabetes as well. The trial found that metformin can lower insulin requirements for type 1 patients. That's a big deal. Metformin has been around for decades, it's inexpensive, and if it can meaningfully reduce how much insulin someone needs, that could change daily life for millions of people managing this condition. More research is needed, but this is a genuinely promising find.

Staying in the world of medicine, scientists have finally cracked open the mechanism behind lecanemab, the FDA-approved Alzheimer's drug marketed as Leqembi. We knew it cleared amyloid plaques from the brain and slowed cognitive decline, but now researchers from VIB and KU Leuven have identified exactly how it does that. The drug works through its so-called Fc fragment, which engages special immune cells in the brain called microglia. These cells then carry out the plaque-clearing work. Understanding the mechanism matters enormously because it opens the door to improving the treatment and developing even better therapies down the line.

And on the aging front, a truly massive new study has mapped nearly seven million cells across twenty-one organs to understand how the body really ages. Scientists found that aging begins earlier than most of us assumed, and it unfolds in a coordinated way throughout the body — it's not just random wear and tear in isolated places. About a quarter of cell types actually change in number over time, and many of these shifts differ between males and females. Crucially, the research highlights shared genetic hotspots that could become targets for future anti-aging therapies. This is one of the most comprehensive cellular atlases ever built, and it's going to fuel aging research for years.

Now, here's a story that caught everyone's attention — and honestly, the headline says it all. Scientists at the University of Maryland have created what they're calling Smart Underwear. Yes, really. It's a wearable sensor system designed to measure human flatulence, and researchers are actively recruiting volunteers for a nationwide study to document the full range of, well, gas output. Look, it sounds like a punchline, but understanding digestive health has real clinical value. Sometimes science takes you to unexpected places, and we respect the commitment.

On the topic of cells and biology, textbooks may need a rewrite. Scientists studying zebrafish embryos discovered that giant embryonic cells can divide in a completely new way. For decades, biology taught us that cells split using a contractile ring — essentially a molecular purse string that cinches the cell in two. But these researchers found that embryonic cells can instead use a mechanical ratchet system to accomplish the same thing. It's an entirely different mechanism, and finding it in a well-studied organism like zebrafish is a reminder that even familiar biology still holds surprises.

Let's talk about superbugs, because this is one of the most pressing health challenges of our time. Drug-resistant bacteria are becoming harder and harder to treat, and the pipeline of new antibiotics has been frustratingly thin. But researchers have now discovered something remarkable: several unrelated viruses have independently evolved a way to disable a key bacterial protein called MurJ, which is essential for building the bacterial cell wall. High-resolution imaging shows that viral proteins lock MurJ into a single position, essentially freezing it — and the bacteria die. This is a natural kill switch, and scientists believe it could inspire a whole new class of antibiotics.

Here's a wild one from quantum computing. Researchers at the Niels Bohr Institute have built an ultra-fast monitoring system that can detect failures in quantum computer qubits in real time — and we're talking milliseconds. Qubits are notoriously fragile; they can flip from stable to unstable in a blink, and that instability is one of the biggest obstacles to building reliable quantum computers. Being able to watch failures as they happen is a critical step toward understanding and eventually preventing them. The team pulled this off using commercially available hardware combined with new measurement strategies, which suggests wider adoption won't be far off.

Now, in the world of artificial intelligence, researchers have unveiled what they're calling Humanity's Last Exam. That sounds alarming, but here's the context: AI systems have been acing standard academic benchmarks — the tests we've traditionally used to measure machine intelligence — and those tests simply aren't hard enough anymore. So scientists have designed a new, far more rigorous benchmark to probe the true limits of what these systems can do. The name is dramatic, but the underlying challenge is a legitimate scientific one: how do we measure machine intelligence when machines keep exceeding our measuring tools?

And speaking of AI, here's a finding that challenges some assumptions about who can spot a fake face. New research shows that the people best at detecting AI-generated faces are not necessarily the most intelligent, nor the most familiar with AI. Instead, it comes down to object recognition skill — the ability to perceive and distinguish fine visual details. If your object recognition is sharp, you're better at catching synthetic faces. Intelligence and AI familiarity, surprisingly, did not predict performance at all. As deepfakes become more common, understanding who's vulnerable and why is increasingly important.

Let's shift to some environmental news, starting with one of those stories that sounds stranger than fiction. Scientists have created seventeen-foot fire tornadoes — intentionally — as a potential tool for cleaning up offshore oil spills. These engineered fire whirls burn oil slicks faster and cleaner than conventional burn methods, producing forty percent less soot and consuming nearly all the fuel. In the critical hours after a spill, responders have to choose between letting oil spread or burning it off. A cleaner, faster burn could make that second option much more viable.

Meanwhile, the coal mining town of Cumberland in British Columbia is testing whether its network of abandoned mines can be repurposed as a town-wide geothermal energy system. Coal mines, it turns out, fill with water over time, and that water retains heat. Tapping into that thermal energy could heat homes and businesses across the entire village at lower cost and with far lower emissions. It's a fascinating example of legacy industrial infrastructure potentially becoming a clean energy resource.

On the topic of plastic, a sobering look at global progress — or lack thereof. The Kunming-Montreal Global Biodiversity Framework set a goal of eliminating plastic pollution by 2030. But meaningful policy action has been slow to materialize. Marine plastic waste doesn't just look bad; it releases PFAS chemicals and heavy metals into the water, doing real biological damage. The science is clear, the goals are set, and yet the gap between ambition and action remains wide. It's a story worth keeping on our radar.

Now for some fascinating evolutionary biology. New research suggests that the earliest animals on Earth had no skeletons at all. By analyzing hundreds of genes and modeling the evolution of skeletal structures, scientists found that mineralized spicules — the hard bits in sponges — arose separately and independently in different sponge lineages. This means the very first sponges were soft and skeleton-free, which also explains why their fossils don't show up until much later in the geological record. It rewrites our understanding of how the first reef-building animals, and possibly the first animals of all, came to be.

And here's a mystery that took two hundred and fifty years to solve. Historical records from eighteenth-century expeditions describe crocodiles as abundant in the Seychelles Islands. They disappeared after human settlement, and no one was quite sure what species they were or where they came from. Now, genetic evidence confirms they were saltwater crocodiles — the same species capable of remarkable long-distance ocean travel. Their historic range apparently spanned over twelve thousand kilometers across the Indian Ocean. How they got there is no longer a mystery: they swam.

Finally, let's end with some surprising fish intelligence. Cleaner wrasse — small reef fish — have long been studied for apparent mirror self-recognition, a trait associated with higher intelligence. But scientists at Osaka Metropolitan University have now observed something even more unexpected. When shown a mirror, these fish didn't just react to their reflection — they began using the mirror strategically, interacting with it in ways that suggest a level of cognitive sophistication previously associated only with mammals. It's another reminder that intelligence in the animal kingdom takes many more forms than we've traditionally assumed.

And that's a wrap on today's episode of Peer Review'd. From smart underwear to fire tornadoes, from ancient soft-bodied sponges to self-aware fish, science continues to deliver the unexpected. Thanks for listening, stay curious, and we'll see you next time.