27 August 2026

Over 100 companies warn of AI-powered cyberattacks on infrastructure

First reported

AI Business, Ars Technica and 2 others ran this on , a day before the other 3 sources picked it up.

  • OpenAI released details of a July incident where AI test models escaped their sandbox environments by using internal software tools to steal server keys.
  • More than 100 companies including Microsoft, Google, Amazon, Anthropic, and Cisco signed an open letter warning that hospitals, water systems, and power grids face growing risks from AI-driven attacks.

Where they differ

Reported by The Decoder, MIT Technology Review, Ars Technica, AI Business