Subscribe
AI News Digest - 2026-09-17
Category
Empty
1.
Ursula von der Leyen warned that AI agents "escaping their environment" were a preview of further risks and said she planned to invite major frontier labs to talks and to use the EU AI Act to help set global AI safety standards, citing autonomous hacking and self‑improving models as immediate risks.
2.
Google DeepMind launched the Deepmind Institute (DMI), an interdisciplinary research platform led by Demis Hassabis, Shane Legg, and James Manyika to study AGI questions including safety, governance, and control risks and to bring arts, humanities, and policy experts together with technologists.
3.
Apple was reported to be building an enterprise AI server using two‑ or four‑chip M8 Ultra configurations for AI inference workloads, with consideration of Nvidia's NVLink Fusion for interconnecting chips and a possible launch no earlier than 2029.
4.
Anthropic merged Claude Chat, Cowork, and related features into a single product in which Claude autonomously selected between quick answers and larger workflows, and the update added Claude Docs and Claude Slides with Pro and Max users receiving first access.
5.
Syensqo described using AI agents and physics‑based simulations to digitally synthesize and rank millions of molecular candidates, which it reported had accelerated materials discovery for semiconductors and data‑center thermal management and supported development of higher‑performance, more sustainable materials.
