AI Presents Escalating Cyber Risks While Driving Regulatory and Defensive Innovations

Key Intelligence

• ■AI is empowering new and scaled cyber threats, including amplified API vulnerabilities, "vibe extortion" by low-skilled criminals, and sophisticated model extraction attacks on LLMs.
• ■Fundamental limitations mean AI models cannot fully ensure security, as demonstrated by product failures like Meta's troubled chatbot for minors.
• ■Governments and regulatory bodies, such as the EU and UK, are responding with new legislation, restrictions on AI use in sensitive environments, and initiatives promoting responsible AI.
• ■Despite the rising threat landscape, AI is also being developed as a defensive tool, learning to autonomously respond to cyberattacks.