OpenAI Reveals Chinese Hackers Used ChatGPT for Cyber Operations
2 min. read 
Published on
OpenAI confirmed a ChatGPT account tied to Chinese law enforcement planned and tracked covert cyberattacks. The group ran “Cyber Special Operations” to harass dissidents and officials worldwide. Details come from OpenAI’s February 2026 threat report.
Hackers used ChatGPT to edit status reports on their campaigns. These covered 300+ social platforms with thousands of fake accounts. Hundreds of operators worked across Chinese provinces. OpenAI banned the account after detection.
One key session targeted Japan’s Prime Minister Sanae Takaichi in October 2025. The user sought help for a smear campaign over her human rights comments. ChatGPT refused but later polished the real attack report. Hashtags and memes spread on X, Pixiv, and Blogspot.
Operation Tactics
The group linked to China’s “Spamouflage” network. They doxxed via revealscum.com. Local AI like DeepSeek-R1 and Qwen2.5 aided monitoring. ChatGPT refined reports.
Targets included activist Li Ying, Safeguard Defenders, and Jie Lijian. Tactics forged U.S. court docs for account bans. Fake obituaries caused distress.
OpenAI’s report states: “This shows industrialized transnational repression from China using AI.”
| Tactic | Target Example | Tools Used |
|---|---|---|
| Influence Campaign | Sanae Takaichi | Fake emails, memes |
| Doxxing/Harassment | Li Ying, Jie Lijian | revealscum.com, fake obits |
| Account Takedowns | Dissidents | Forged U.S. court docs |
| Monitoring | Social posts | DeepSeek-R1, YOLOv8 |
Over 100 tactics aimed at silencing critics.
Broader Impact
Operations hit foreign officials and rights groups. They mixed AI content with human ops. Platforms saw mass reports with fake proof.
This ties to past Spamouflage exposes by Meta. OpenAI shared intel to boost defenses. State actors evolve fast with AI.
Defense Steps
Platforms and users can act now:
- Detect coordinated fake accounts and AI reports.
- Verify legal docs before takedowns.
- Monitor for mass harassment patterns.
- Share threat intel across firms.
Governments warn activists. AI firms tighten policies.
FAQ
Chinese law enforcement-linked group in “Cyber Special Operations.”
Editing status reports on harassment campaigns.
No. It refused planning but polished post-attack docs.
Dissidents like Li Ying, Japan’s Sanae Takaichi.
Watch for AI-generated fakes and mass reports.
Read our disclosure page to find out how can you help VPNCentral sustain the editorial team Read more
Improve this guide
User forum
0 messages