Anthropic report: Chinese labs mined Claude for 151M+ exchanges; abuse tied to missiles, drones, surveillance

The Decoder · rss · 2026-09-11

Anthropic's new threat intelligence report documents eight months of Claude abuse. Chinese AI labs — including Alibaba's Qwen team, DeepSeek, and Moonshot AI — allegedly relayed requests en masse or extracted training data, with Qwen alone linked to more than 151 million exchanges. The report also details actors using Claude for missile software, autonomous kamikaze drones, and nationwide surveillance systems.

Original post →

More from Safety

Safety channel →