A 48-year-old man with a neuroendocrine tumor walked into a PET scan in Kraków, Poland, and left with what sounds like science fiction. Not because of what the scan found in his pancreas, but because for the first time, his doctors could see quantum entanglement unfolding inside his living body.

For years, decoding brain activity meant picking one direction: you could either read what someone was seeing from their fMRI scan, or you could predict brain activity from an image. Never both in the same system. A team of researchers from Tianjin University, Shanghai Artificial Intelligence Laboratory, and the Chinese University of Hong Kong just tore down that wall. Their model, BrainJanus, reads brain signals and generates them, all within a single framework.

In 1974, a group of scientists at Imperial College discovered that when you shine light on a rough silver surface, the molecules sitting on top of it scatter far more light than they should, about a million times more. That accidental finding, surface-enhanced Raman spectroscopy (SERS), gave chemists a superpower: the ability to identify individual molecules by the way they vibrate. A few years later, researchers found that infrared light, the kind your TV remote uses, could be similarly amplified on nanostructured surfaces. The catch? Nobody could get both techniques to work on the same device.

In June 2026, a 62-year-old melanoma patient who had already failed two lines of therapy received an experimental vaccine. It wasn’t an off-the-shelf drug. It was built from the DNA of their own tumor, encoding 34 custom neoantigens selected by an algorithm trained on thousands of other patients’ cancers. Six months later, scans showed no new lesions.

That vaccine, Moderna’s mRNA-4157 (now called V940), is the most visible example of a quiet revolution in the open-source code repositories where computational biologists build the tools that make personalized cancer immunotherapy possible. One of those tools, pVACtools, just got its biggest update yet.

In 2025, IonQ unveiled a 98-qubit trapped-ion quantum computer with all-to-all connectivity, a machine that could run algorithms no classical computer could simulate. It was a genuine milestone. But even that machine had a fundamental constraint: its ions were held in place by voltages applied to distant metal electrodes, and moving them required nudging them through a static maze. The parallel that quantum computing researchers have been chasing, the ability to reconfigure hundreds of qubits as easily as rearranging pins on a circuit board, remained elusive.

Now a team from the Max Planck Institute for Quantum Optics, Duke University, the University of Innsbruck, and IonQ believes they’ve found a way around that limitation. Their proposal marries two of the most successful quantum computing platforms, trapped ions and optical tweezer arrays, into a single architecture that could finally give engineers the flexibility they need to scale up.

1|In 2020, a team at MIT used reinforcement learning to discover halicin, a powerful new antibiotic hiding in plain sight among thousands of existing drugs. The AI found it in days. What most people missed was the bottleneck that makes that kind of discovery so rare: before you can search for new molecules, you need to understand how existing ones behave, and that means simulating them at the atomic level. That computational nightmare quietly strangles drug discovery, materials science, and every field that depends on molecular simulation. 2| 3|

Every year, millions of parents-to-be stare at ultrasound images, hoping for reassurance and bracing for uncertainty. For about 6% of all pregnancies worldwide, the scan reveals something unexpected: a structural anomaly in the developing fetus. A heart that formed differently. A spine that didn’t close. In the moments that follow, a grueling diagnostic odyssey begins — one that, even with the best genetic sequencing technology, can take months or years and often ends without answers. A new artificial intelligence system from Zhejiang University, described in a preprint posted this week, just made that journey dramatically shorter. Its name is DeepBD, and it might be the closest thing yet to a diagnostic safety net for the most vulnerable patients who will never speak for themselves.

Deep in the laboratory of the University of South Bohemia, a team of scientists has done something that would have sounded like science fiction just a decade ago: they took a virus that infects bacteria, hollowed it out, stuffed it with a photosynthetic protein and an electron-carrying partner, and turned the whole assembly into a working solar-powered nanoreactor. The result, published this week on arXiv, is a modular molecular toolkit that could rewrite how we think about artificial photosynthesis — and eventually, how we produce fuels, chemicals, and medicines using nothing but sunlight.

If you could board a spaceship capable of flying in a perfectly straight line forever, what would happen? Common sense says you’d travel through endless galaxies, never reaching an edge, never looping back. But common sense might be wrong. The universe could be finite — and if you flew far enough, you might eventually return to where you started, having circumnavigated all of existence like a cosmic Magellan. The question of whether the universe wraps back on itself — its topology — is one of the deepest and most persistently overlooked questions in cosmology. Now, a comprehensive review from the COMPACT Collaboration, invited for publication in Nature Astronomy, lays out everything we know, everything we don’t, and the astonishing possibility that the evidence may already be hiding in data we’ve already collected.

Every time you ask ChatGPT or Gemini a hard question, something remarkable happens behind the scenes. The model doesn’t just blurt out an answer — it thinks. It writes out a stream-of-consciousness chain of reasoning, step by step, in plain English, before delivering its final response. This isn’t just a quirk; it’s a safety feature. When an AI writes down its thoughts in human language, researchers can read them. They can spot signs of deception, catch flawed logic, and — if the model ever starts plotting something dangerous — hopefully intercept it before it’s too late.