Research

Anthropic reported on September 29 that Z.ai’s open-weight GLM-5.3 model could build working browser exploits and that researchers could substantially weaken its refusal safeguards. The finding concerns a model that anyone can download, modify and run. Anthropic’s Frontier Red Team tested whether GLM-5.3 could turn known software defects into working attacks and whether it would follow explicitly harmful instructions after common safeguard-bypass techniques. Anthropic researchers Andrew Fasano, Marius Fleischer, Cole McFaul, Robert Xiao and Tripp Gallagher authored the report. The team ran models in isolated environments and combined automated benchmarks with sessions in which security researchers directed the model while examining unfamiliar software targets. ...

A British Transport Police facial-recognition trial scanned more than half a million faces but produced one incorrect watchlist alert and no arrests caused by an alert, the Guardian reported on September 29. The six-month pilot covered 18 deployments at busy London railway stations between February and July. Records obtained through a freedom-of-information request put equipment and staffing costs at £320,786 and police time at almost 100 hours. Why it matters: Rail passengers had their biometric data processed at scale, while the system generated no correct watchlist match during the reported period. That result gives lawmakers and oversight bodies a concrete deployment record for judging whether the intrusion was proportionate. ...

PostHog released a reasoning-based decision model while OpenAI expanded its developer, pricing and enterprise-distribution offers. Community discussions focused on privacy measurements, data-centre economics and possible restrictions on Chinese open weights. » Why it matters: The day’s lower-ranked releases and discussions show where AI products are being packaged for routine work—and where users still lack independent evidence about performance, privacy and cost. In brief PostHog adds reasoning to a decision model. Jeeves is a 9-billion-parameter, Jev-compatible model that answers yes-or-no, multiple-choice and rating questions after an optional reasoning stage. PostHog released weights, code and data and reports stronger results than Jev on selected public tests; those results remain developer-run. It is worth watching because it combines a constrained decision interface with a slower reasoning path instead of sending every classification task to a general chatbot. Source ...