Researchers planted fake research for $14; Doubao cited it within 90 minutes
Sanbi_Ai · reddit · 2026-10-15
A Fudan University paper, "From Public Posts to AI-Search Citations" (accepted at NDSS 2027), shows how cheap it is to poison AI search answers:
- Researchers invented a research technique that doesn't exist, complete with fake acronyms and stats, and paid a Chinese "get your brand into AI answers" vendor $14. Within 28 minutes, 13 posts were live on Sohu, Toutiao and Zhihu; within the hour, Doubao — ByteDance's chatbot with 382M monthly users — was citing one and repeating the planted details.
- Controlled tests: publishing a fake concept on sites AI engines already like to cite got 8 of 10 platforms citing it within a week, including ChatGPT, Google AI, Perplexity and Grok. Doubao picked it up day one and kept citing posts after deletion.
- Where you post matters far more than volume: 40 posts on a self-hosted WordPress blog got zero citations in a week.
- DeepSeek was the only platform that never cited any planted content in the main tests.
- Surprise finding: Chinese chatbots lean hard on parent-company sites — 25.9% of Doubao's citations went to ByteDance-owned domains, 13.5x their share elsewhere; Baidu's Wenxin (13.7x) and Tencent's Yuanbao (6.3x) behaved similarly.
Caveats: zero-competition fake topics are the easiest case, numbers are a spring-2026 snapshot, and it doesn't mean cited pages are paid; all findings were reported to all 10 platforms.
More from Safety
- EU publishes its first AI standard: EN 18286 on AI quality management — LuizaJarovsky · 2026-10-15
- Korea clears 20 humanoid robots to work on real factory floors — emmanuelvivier · 2026-10-15
- Microsoft locks down Copilot against prompt injection with deterministic isolation layer — emmanuelvivier · 2026-10-15
- Gladstone AI report: labs automate R&D pipelines, RSI could start in 3-12 months — sourdub · 2026-10-15
- Hackathon team claims it found a large fleet of agents from a non-US frontier lab — NathanpmYoung · 2026-10-15
- Argon hits cyber, CBRN CCLs and misalignment TCL; report flags CoT transparency — vkrakovna · 2026-10-15