How much have AI applications learned, and how can one know their capabilities if they are being evaluated with an exam that is far too easy? In 2024, with the publication of the previous benchmark to measure AI performance, the field debated whether existing assessments would keep pace with the rapid progress of AI.

Google Deepmind is shedding light on the dark genome with its latest AI model, which is trained to decipher the 98% of DNA that does not code for proteins. Alphagenome is designed to predict how variants in the regulatory genome exert their effects on the expression of the genes they control.

Zylox-Tonbridge Medical Technology Co. Ltd. plans to expand globally with M&As, company chairman and CEO Jonathon Zhong Zhao told BioWorld, starting with a stepwise acquisition of Optimed Holding GmbH and its subsidiaries, including Optimed Medizinische Instrumente GmbH.

Beta Bionics Inc.’s preliminary results for its first full year as a public company offered an early read on developments and market positioning in the insulin pump and patch market, with analysts pointing to steady growth, rising pharmacy channel penetration and intensifying pressure as more products target both type 1 and type 2 diabetes.

Liver fibrosis in the course of metabolic dysfunction-associated steatohepatitis (MASH) could be significantly reduced using CAR T-cells generated in vivo. Scientists at the Icahn School of Medicine at Mount Sinai have developed an experimental cell therapy that eliminates only one type of liver cell, the stellate cells that express fibroblast activation protein alpha (FAP). This strategy not only reduced fibrosis but also reversed liver damage.

A new method, based on gene editing with oligonucleotides and functional analyses, identifies which variants of DNA repair genes associated with Lynch syndrome are truly harmful and which are not. Scientists at The Netherlands Cancer Institute have developed this technique and classified these gene variants in both coding and noncoding regions, distinguishing those that are pathogenic from those that are benign.

Depending on who you ask, AI will take over the world and save it; or ruin it. Certainly, it is changing it. Science magazine dedicated its first editorial of 2026 to AI. Despite its title – “Resisting AI slop“ – editor-in-chief Holden Thorp gave the sort of nuanced review that is typical of him.

Depending on who you ask, AI will take over the world and save it; or ruin it. Certainly, it is changing it. Science magazine dedicated its first editorial of 2026 to AI. Despite its title – “Resisting AI slop“ – editor-in-chief Holden Thorp gave the sort of nuanced review that is typical of him. “Like many tools, AI will allow the scientific community to do more if it picks the right ways to use it,” he wrote. “The community needs to be careful and not be swept up by the hype surrounding every AI product.”

The concept of the 3 Rs – reducing, refining and replacing animal research – has been championed since the 1950s, when William Russel and Rex Burch argued in their book “The Principles of Humane Experimental Technique” that the 3 Rs could simultaneously improve the treatment of research animals and advance the quality of scientific and medical research and testing. Current standard practices of animal research undeniably cause animal suffering at the same time that they have prioritized replicability over translatability.