Security · 1h ago
AI worms can spread via Microsoft Copilot for Word
Researchers demonstrate document-borne AI worms that self-propagate through Microsoft Copilot for Word. The worms embed adversarial prompts in documents, tricking the AI into executing malicious actions. This highlights new security risks as AI assistants gain access to sensitive data and systems.
Meridian48 take
While the proof-of-concept is alarming, real-world exploitation depends on specific configurations and user permissions, but it underscores the need for robust AI security measures.
Read the full reporting
Document-borne AI worms can self-propagate through Copilot for Word →
Hacker News
ai-wormscopilot-security