New research from Cornell University indicates that AI agents powering tools like ChatGPT and Google's AI search can be easily manipulated by short, user-generated text snippets. The study, titled "Deep-research agents can be poisoned via user-generated content," demonstrates that brands can inject promotional content into sites such as Reddit, Quora, and Wikipedia to influence AI outputs. This practice, known as AI-engine optimization (AEO), is reportedly flooding these platforms with inauthentic content. AI
IMPACT AI models are susceptible to manipulation through user-generated content, potentially impacting search results and AI-powered tools.
RANK_REASON Research paper detailing a new vulnerability in AI models. [lever_c_demoted from research: ic=1 ai=1.0]
Read on Mastodon — mastodon.social →
- ChatGPT
- Cornell University
- Hal Triedman
- Mastodon
- Quora
- Tingwei Zhang
- Vitaly Shmatikov
- Wikipedia
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →