A redesigned cancer immunotherapy is showing striking early results after decades of disappointment with similar drugs. Researchers engineered a more powerful CD40 agonist antibody and changed how it’s delivered—injecting it directly into tumors instead of into the bloodstream. In a small clinical trial of 12 patients with metastatic cancers, six saw their tumors shrink and two experienced complete remission.

A new UCLA Health study suggests that long-term exposure to the pesticide chlorpyrifos may dramatically raise the risk of Parkinson’s disease. Researchers found that people living in areas with sustained exposure had more than 2.5 times the likelihood of developing the disorder. Lab experiments reinforced the finding: animals exposed to the chemical developed movement problems, lost dopamine-producing neurons, and showed the same toxic protein buildup seen in Parkinson’s patients.

Background: “I’m not a doctor, but...” is a typical response when asking considerate laypeople for health advice. However, seeking medical advice has also shifted to digital settings, where the expertise of the other party is less transparent than in face-to-face interactions. Recently, large language models (LLMs) have emerged as easily accessible tools, offering a novel way to formulate medical questions and receive seemingly qualified advice. Given the sensitive nature of health-related queries and the lack of professional supervision, incorrect advice can pose serious health risks. Therefore, including explicit disclaimers and precise referrals in LLM responses to medical queries is crucial. However, little is known about how LLMs adapt their safety implementations in response to different urgency levels. Objective: This study evaluates disclaimer and referral patterns in responses from LLMs to authentic medical queries of different urgency levels using a systematic evaluation framework. Methods: This prospective, multimodel evaluation study generated and analyzed 908 responses from 4 popular LLMs (GPT-4o, Claude Sonnet-4, Grok-3, and DeepSeek-V3) to 227 authentic patient queries from a public dataset. Two human raters classified all 227 patient queries using a 3-level urgency scale. LLM responses were evaluated using a 5-point ordinal classification system for disclaimer and referral advice, ranging from “no disclaimer” to “urgent advice to consult a medical professional.” GPT-4o served as the primary rater model for this task after conducting a subset validation against human expert annotations. Statistical analyses included Jonckheere-Terpstra tests to examine the relationship between case urgency and disclaimer ratings and Kruskal-Wallis tests for intermodel comparisons. Results: The 227 patient queries were distributed as 77 (34%) low-urgency, 110 (48%) intermediate-urgency, and 40 (18%) high-urgency cases. All 4 LLMs demonstrated statistically significant ordered trends (all

A clinical trial found that listening to specially designed music with auditory beat stimulation can significantly reduce anxiety. Among several listening lengths tested, a 24-minute session delivered the biggest benefits, easing both mental and physical symptoms of anxiety. The results suggest there may be an ideal “dose” of therapeutic music that works quickly without requiring long listening sessions.

New research shows that sourdough fermentation does more than make bread rise—it transforms wheat fibers in unexpected ways. Scientists found that enzymes already present in wheat, activated by the sourdough’s acidic environment, break down key fibres called arabinoxylans. This process may influence the bread’s texture, digestibility, and flavor. Some microbes even create buttery or subtly sweet notes that give sourdough its distinctive taste.

The distinctive smell of ancient mummies is helping scientists decode the secrets of Egyptian mummification. By analyzing tiny traces of chemicals in the air around mummy samples, researchers identified dozens of compounds linked to oils, resins, beeswax, and bitumen used during embalming. The chemical clues reveal that mummification grew increasingly complex over thousands of years.

Physicists at UC Santa Barbara have uncovered a new way to manipulate unusual magnetic states by exploiting “frustration” inside a crystal’s atomic structure. The team discovered a rare system where two different kinds of frustration—magnetic and electronic bond frustration—coexist and interact. By coupling these competing effects, researchers may be able to control exotic quantum states, potentially unlocking new ways to manipulate entangled spins for future quantum technologies.

Scientists have developed a powerful new computational method that could accelerate the search for next-generation materials capable of turning sunlight into useful chemical energy. The work focuses on polyheptazine imides, a promising class of carbon nitride materials that absorb visible light and can drive reactions such as hydrogen production, carbon dioxide conversion, and hydrogen peroxide synthesis. By analyzing how 53 different metal ions influence the structure and electronic behavior of these materials, researchers created a framework that predicts which combinations will perform best.

Scientists have discovered that cells can sense far beyond the surfaces they touch. While individual cancer cells can probe about 10 microns ahead by tugging on surrounding collagen fibers, clusters of normal epithelial cells can combine forces to detect layers as far as 100 microns away. This long-range “depth sensing” helps cells decide where to migrate. Understanding how it works could reveal new targets to stop cancer from spreading.

Researchers have developed a tiny antibody that can find a common cancer protein and make tumors light up during PET scans. In tests with mice, tumors containing the protein EphA2 glowed clearly when the antibody was used. This could help doctors quickly identify patients who may benefit from new targeted cancer treatments. The method may also provide a faster and less invasive alternative to traditional testing.

Scientists at MIT have discovered that a little-known protein called intelectin-2 plays a powerful double role in defending the gut. The protein strengthens the mucus layer that lines the gastrointestinal tract while also trapping and disabling harmful bacteria that try to break through. By binding to sugars on both mucus molecules and bacterial surfaces, intelectin-2 forms a protective barrier and can even destroy microbes, including some that resist antibiotics.

Artificial intelligence is often portrayed as a tool that replaces human work, but new research from Swansea University suggests a far more exciting role: creative collaborator. In a large study with more than 800 participants designing virtual cars, researchers found that AI-generated design galleries sparked deeper engagement, longer exploration, and better results.

A new AI framework called THOR is transforming how scientists calculate the behavior of atoms inside materials. Instead of relying on slow simulations that take weeks of supercomputer time, the system uses tensor network mathematics and machine-learning models to solve the problem directly. The approach can compute key thermodynamic properties hundreds of times faster while preserving accuracy. Researchers say this could accelerate discoveries in materials science, physics, and chemistry.

Nature, Published online: 16 March 2026; doi:10.1038/s41586-026-10372-3

Direct conversion from alkenes to alkynes

Nature, Published online: 16 March 2026; doi:10.1038/d41586-026-00769-5

Microbial diversity loss seen after a course of some commonly prescribed antibiotics can persist for years.

Nature, Published online: 16 March 2026; doi:10.1038/s41586-026-10179-2

A machine-learning model that integrates data from wearable devices (such as smartwatches) with blood biomarkers and demographic data can predict whether someone has insulin resistance, enabling timely lifestyle interventions to prevent progression to type 2 diabetes.

Nature, Published online: 16 March 2026; doi:10.1038/d41586-026-00843-y

Graduate students increasingly use artificial-intelligence tools to draft, code and search — but many fear it could erode the very skills a doctorate is meant to build.