Many security leaders didn’t authorize AI expansion. It happened around them. Someone plugged in a copilot in a SaaS tool or an engineering team tested an agent or a business unit installed an assistant without waiting for approval. None of these choices feel significant in isolation, but together they create systems acting on behalf of people, without the structures we rely on to govern human access.
In our survey of more than 200 CISOs and security leaders, the same concerns surfaced repeatedly. AI systems already have meaningful access, often with privilege levels no one explicitly granted. They generate activity that can be difficult to trace, behave in ways that don’t match human patterns, and sometimes leave behind incomplete or temporary records. None of this is catastrophic on its own, but it complicates the basic questions security teams rely on, namely: “Who did this?” and “Should this action have been allowed?”
Leadership teams are worried because AI is already reading customer data, modifying configurations, invoking APIs, and chaining actions together in ways that are difficult to trace back to a single owner. AI identities don’t behave like human users or traditional service accounts.
Security leaders are clear-eyed about the challenge. They want workable visibility, a way to understand how these systems operate, and a practical path to keep privileges from quietly expanding beyond what anyone intended. This report focuses on what leaders are dealing with right now. AI is active in production environments, and most organizations can’t clearly explain the scope of its access.
Learn more about the report and its key findings by downloading the full report using the button to the right.
