Story perspectives

Kremlin-Backed Group Used Claude to Program Lethal Drones

By Drooid · Sep 15, 2026, 3:37 AM UTC · How we work

Summary

Story summary

Kremlin-backed actors used Anthropic's Claude AI to program lethal drones that select and detonate targets.
GTG-27005 built and hardware-in-loop tested a Claude-based model for single-board computers, but did not deploy.
Anthropic banned GTG-27005, added safeguards, and shared details with industry groups.
Anthropic disrupted Chinese and Yemeni groups attempting to design conventional weapons with Claude.
Foreign actors used Claude to steal data, create malware, forge documents, and generate fake NATO-Baltics narratives.

1 of 1

Story summary

Kremlin-backed actors used Anthropic's Claude AI to program lethal drones that select and detonate targets.
GTG-27005 built and hardware-in-loop tested a Claude-based model for single-board computers, but did not deploy.
Anthropic banned GTG-27005, added safeguards, and shared details with industry groups.
Anthropic disrupted Chinese and Yemeni groups attempting to design conventional weapons with Claude.
Foreign actors used Claude to steal data, create malware, forge documents, and generate fake NATO-Baltics narratives.