Scientists from major research centres have issued a stark warning: current large language models (LLMs) driving AI-powered robots are not yet safe for unsupervised use in homes, care settings, or other spaces where physical harm and discrimination can occur. The peer‑reviewed study published in...