CyberHappenings logo
☰
Home Browse Watchlists Experimental About

Track cybersecurity events as they unfold. Sourced timelines. Filter, sort, and browse. Fast, privacy‑respecting. No invasive ads, no tracking.

AI Recommendation Poisoning via 'Summarize with AI' Buttons

First reported
Last updated
1 unique sources, 1 articles

Summary

Hide ▲

Microsoft has identified a new AI manipulation technique called AI Recommendation Poisoning, where businesses use 'Summarize with AI' buttons to inject biased recommendations into AI chatbots. This method involves embedding hidden instructions in URLs to manipulate the AI's memory and skew recommendations. Over 50 unique prompts from 31 companies across 14 industries were detected over a 60-day period. The attack leverages the AI's inability to distinguish genuine preferences from injected ones, potentially leading to biased advice on critical subjects like health, finance, and security.

Timeline

  1. 17.02.2026 11:31 1 articles · 13h ago

    Microsoft Identifies AI Recommendation Poisoning via 'Summarize with AI' Buttons

    Microsoft has uncovered a new AI manipulation technique where businesses use 'Summarize with AI' buttons to inject biased recommendations into AI chatbots. This method involves embedding hidden instructions in URLs to manipulate the AI's memory and skew recommendations. Over 50 unique prompts from 31 companies across 14 industries were detected over a 60-day period. The attack leverages the AI's inability to distinguish genuine preferences from injected ones, potentially leading to biased advice on critical subjects like health, finance, and security.

    Show sources
    Open in new tab

Information Snippets