LINKDING

 Shared bookmarks

Shared bookmarks

  • Evaluating Malicious Generative AI Capabilities
    #AI #AIrisks #generativeAI | This CETaS Briefing Paper examines how Generative AI (GenAI) systems may enhance malicious actors’ capabilities in generating malicious code, radicalisation, and weapon instruction and attack planning. It synthesises insights from government practitioners and literature to forecast inflection points where GenAI could significantly increase societal risks. Unlike previous studies focused on technological change, this paper also considers malicious actors’ readiness to adopt technology and systemic factors shaping the broader context. It argues for a sociotechnical approach to AI system evaluation, urging national security and law enforcement to understand how malicious actors' methods may evolve with GenAI
    4 months ago | View Shared by teclista

User


Tags

-AIrisks -generativeAI

AI