Microsoft discovers "AI Recommendation Poisoning" hidden in Summarize buttons across 31 companies

Microsoft's Defender Security Research Team found that dozens of companies embed hidden prompt injections inside innocent-looking "Summarize with AI" buttons. Clicking one opens an AI assistant with a pre-loaded manipulative instruction that permanently skews its recommendations toward that company. Researchers identified over 50 manipulative prompts from 31 companies across 14 industries in just 60 days. Freely available website plugins make the technique accessible to anyone. Microsoft recommends verifying target URLs and regularly deleting AI assistant memory.

View full digest for February 22, 2026