WorkSeptember 11, 2026via The Decoder
How hackers used Claude for missiles, drone swarms, and surveillance, while Chinese labs mined it for training data
Why it matters
Claude abuse at scale — both state-adjacent and weaponized — forces a reckoning on AI safety, data exfiltration, and how frontier labs monitor misuse. This is the first concrete evidence of systematic exploitation by competing labs and the policy/security conversation that follows.
Key signals
- Anthropic released eight-month threat intelligence report on Claude abuse
- Qwen (Alibaba) made 151+ million requests to Claude, data-mining for training
- DeepSeek and Moonshot AI also documented as relaying requests en masse or extracting training data
- Documented misuse cases: missile software, autonomous kamikaze drones, nationwide surveillance systems
- Evidence of state-adjacent and weaponized AI model use
The hook
151 million Claude queries. Chinese labs mining Anthropic for training data while bad actors built missiles and drone swarms on the platform.
Anthropic's new threat intelligence report documents eight months of Claude abuse. Chinese AI labs like Alibaba's Qwen team, DeepSeek, and Moonshot AI relayed requests en masse or extracted training data, with Qwen alone accounting for more than 151 million exchanges. Actors also used Claude for mis…