AI Security Incidents Surge: Adoption Outpaces Readiness

Organizations worldwide are seeing a surge in AI-related security incidents. More than eight out of ten companies experienced at least one AI security breach in the past year. This shows that AI adoption is happening faster than companies are ready for it. A survey of 750 respondents found that nearly 40% reported prompt injection attacks. Over 20% detected non-approved generative AI assistants, and 18% had an AI assistant leak sensitive data. Also, over 4% had unauthorized access to their genAI tools. Data leakage was the most common breach, cited by over 50% of respondents. Manipulation of AI agents by malicious input followed closely, at nearly 50%. What's interesting is that 76% of professionals rank data security and privacy as their top concern with AI. These risks are even delaying AI implementation, with 86% of organizations postponing it by an average of almost six months. This highlights how data security is a major barrier to AI adoption, even more than technical limits or regulations. To address these issues, nearly 95% of organizations have taken at least one action in the last year. Adding human oversight for AI agents is the most common strategy, used by almost 55% of respondents. The bottom line is that while AI offers many benefits, its rapid adoption without proper security measures creates significant risks that organizations are now grappling with.