PromptSpy: First Android Malware to Embed Generative AI for Adaptive Behavior
Published 2026-02-25AI Regulation and GovernanceMedium⭐ Timeline Candidate
Summary
Security researchers have identified PromptSpy, described as the first Android malware to harness generative AI — specifically Google's Gemini model — to make adaptive decisions about its behavior. Rather than following hardcoded instructions, the trojan queries a language model to determine context-aware actions, representing a significant evolution in malware sophistication. The discovery follows ESET's August 2025 research into PromptLock, which was characterized as the first AI-driven ransom
Radar Context
Cursor
Alignment: New signal not yet covered
ai-malwarepromptspyandroid-securitygenerative-ai-threatsgeminiai-governancecybersecurityagentic-malwareesetai-safety