Security Pros Reckon the Biggest AI Risk Comes With an Employee Badge
Help Net Security, Monday, October 26th, 2026
An O'Reilly survey finds security pros worry more about employees leaking data into AI tools than about AI-powered attacks.
An O'Reilly survey of 209 security professionals found that data leaking through commercial AI tools and staff trusting unchecked AI output were each top concerns for 49% of respondents, while AI-generated phishing worried only about a quarter despite 55% encountering it in the past year.
Over half said coworkers at least occasionally put sensitive data into unapproved AI tools, and shadow AI was less widespread at firms with comprehensive policies.
Respondents' top AI agent worry was excessive permissions or unclear ownership.
Two-thirds of defenders use chatbots or copilots for research, and only 7% reported no skills gap.