threatintel
actor tracker
AI misuse catalog

Spamouflage Chinese influence operation used LLMs to draft posts

2024-05-30OpenAIChatGPT
CN · China
ChinaVendor label:Spamouflage
Influence operationsScripting / automation
Primary disclosure (OpenAI)

What they were trying to do

  1. 01

    Generate short social-media comments in multiple languages for cross-platform distribution

  2. 02

    Debug and improve scripts automating coordinated posting across social platforms

How they did it

Spamouflage used OpenAI models to generate short multilingual social-media comments for cross-platform distribution and to debug automation scripts used in the posting workflow. The operation bolted AI into an existing scripted distribution infrastructure rather than replacing it.

Example queries

Verbatim excerpts from the disclosure — only what the vendor actually published.

Not detailed in the disclosure

What they achieved

Concrete outputs obtained before platform disruption.

  • before disruption

    Short social-media comments in multiple languages generated for distribution

  • before disruption

    Debugging assistance for cross-platform posting automation scripts

Vendor responseOpenAI

OpenAI banned the associated accounts. OpenAI assessed the operation had not succeeded in meaningfully building audience engagement before disruption.

Full summary

PRC-linked Spamouflage (DragonBridge) operation used OpenAI models to generate short social-media comments in multiple languages and to debug scripts that automated cross-platform posting. OpenAI banned the accounts.

AI platforms involved

ChatGPT

Related incidents (same country or vendor label)

Source note: Every entry in this catalog is derived from a primary public disclosure by the named platform vendor. This entry cites OpenAI as the primary source. No private, speculative, or non-public claims are presented here.