Anthropic is putting its AI directly into the hands of the companies that keep the lights and water on.
The company announced the Critical Infrastructure Defense Program on Thursday, the first initiative under a broader effort it calls the Anthropic Cyber Mission. Eleven partners signed on, including CrowdStrike and Hitachi. Anthropic says it will give these firms access to its frontier Claude models, its own engineers and in-house threat research focused on power grids, water utilities and transport networks. The program is framed as a long-term commitment rather than a one-off pilot.
Critical infrastructure runs on decades-old industrial control systems that were never built with modern cyberattacks in mind, and the operators who run them are chronically short on security staff. Pairing that gap with a well-funded AI lab's threat-research team is a bigger bet than another chatbot feature - it is Anthropic trying to become infrastructure for the infrastructure defenders.
CrowdStrike already sells its own AI-assisted detection tools, so this is as much a distribution deal for Claude as it is a public-safety initiative - worth watching whether the threat research holds up once attackers actually start testing it.