100+ organizations warn AI attacks will scale

OpenAI says more than 100 organizations have signed its collective cyber-defense letter, warning that increasingly capable AI could make cyberattacks more widespread and sophisticated in the coming months. Reporting on the signatories names Anthropic, AWS, Google, Microsoft, Cisco, IBM, and CrowdStrike, but the document is a call to action rather than evidence of a single new breach. The operator takeaway is to treat model-enabled attack scale as a patching and access-control deadline, not a future scenario; Microsoft’s 90%-small-model security routing shows the defensive side of the same shift.