Researchers at Columbia University (Cornish lab and Ken Shepard lab) used an AI design-build-test cycle to engineer yeast that senses PFOA and PFOS, and built a dip-in test strip that reads PFAS levels in water instantly.
Haiqu has launched AgenticOS, an AI agent system built for scientific discovery that uses cross-validation among multiple teams to stop general-purpose agents from quietly altering the science. Used in high-temperature superconductivity research on the Hubbard model, it cut circuit depth by 75%.
An unreleased research version of Claude has raised a zeta bound related to the Riemann hypothesis from 41.6% to 67.2%, according to a post on Anthropic's research site on October 5 that Ground News aggregated on October 7.
Six cognitive scientists have published a formal proof called the "Ingenia theorem": even under ideal conditions with perfect data, achieving human-level AI through learning remains an NP-hard problem, undermining the narrative that scaling inevitably leads to AGI.
A team combined AI protein structure prediction with high-throughput proteomics to discover molecular-glue degraders that target VAV1, a key immune-signaling protein, as reported by Hanfeng Lin et al. in Nature Communications in 2026.
A team at LMU Munich has developed SPARCS, which uses AI to analyze microscope images of 70 million cells at once and systematically screen how gene changes affect cells, as reported in Cell on October 8, 2026.
Butterfly Effect, the parent company of Manus, has raised more than $500 million in a new round led by Boyu Capital and IDG Capital, with existing investors Tencent, Sequoia China and ZhenFund adding more, making it one of the largest AI agent financings this year.
From October 9, Google is restricting model access in Gemini: free users get only the weakest model, Flash-Lite, while the Flash and Pro models they could previously choose are withdrawn.
Anthropic says its Claude model helped discover a new enzyme system in bacteriophages, array-associated reverse transcriptases (ART), which feature CRISPR-like repeat arrays; 950 agents found it in 21 hours.
Illumina released SpliceAI2 on October 8, a genomic AI model that predicts how genetic variants alter RNA splicing, a key step in interpreting variants of uncertain significance.
XtalPi disclosed details of its Kodexia platform on October 6, in which generative AI, biological modeling and automated experiments form a closed loop to speed up the discovery of small interfering RNA (siRNA) therapies.
A New York University team reports in Chemical Science an AI trained to predict hydrogen atom positions in drug molecules from stability-related chemical patterns, tackling the long-standing problem of tautomer assignment.
AI drug discovery company Transcripta Bio and "AI chemistry foundry" Synfini report obtaining a validation-grade lead compound against the MSH3 gene in a matter of weeks, versus an industry benchmark of 12-18 months from hit to candidate nomination.
On October 8, the US Department of Energy announced 12 Phase II projects under the Genesis Mission, totaling $159 million, to embed AI deeply in the scientific discovery workflows of the national laboratories.
Francis Halzen, winner of the 2026 Nobel Prize in Physics, says the IceCube neutrino observatory solved "a very interesting problem for AI", and that AI played a pioneering role.