news
El Financiero
2026-09-11
Cybercriminals Use Indirect Prompt Injection to Deceive AI Systems
Researchers have identified a new threat where hackers embed hidden instructions into applications to manipulate how AI models interpret code and avoid security detection.