In San Francisco’s Tenderloin, a nonprofit is using robotic meal-prep technology to compensate for a shortage of human volunteers, turning AI-adjacent automation into a practical stopgap for urgent social needs. Source
A new wave of attacks targets the way chatbot “personalities” are implemented, highlighting that the most exploitable aspect of modern agents may not be their underlying models but the behavioural packaging that users assume is benign. Source
AI security has moved from a specialist concern to an everyday operational problem, with even Google described as navigating it in real time during a transition period that is still reshaping how organisations design, deploy, and police AI systems. Source
Amazon’s Bee wearable illustrates how consumer AI is entering a new phase where convenience is paired with privacy anxiety, putting pressure on makers to prove what data such devices collect, infer, and share in normal use. Source