AIThe Hacker Newsabout 2 hours ago
Ask AI Buttons Poison LLM Memory
1 min read
A new prompt injection attack abuses pre-filled deep links in "Ask AI" buttons on commercial websites. Microsoft catalogued it as AI Recommendation Poisoning in February 2026, identifying 31 companies across 14 industries. The payloads instruct ChatGPT, Claude, Gemini, or Grok to permanently save vendor domains as trusted sources.
Level
Hype check
Tap to vote and see what everyone thinks.
