AI is advancing.
How must assurance keep pace?

Greater capability. A more complex basis for confidence.

AI is expanding what organisations can analyse, generate and delegate. It also creates opportunities to examine more evidence and strengthen assurance. At the same time, models, tools and increasingly autonomous agents introduce new combinations of behaviour, dependencies and authority that conventional testing may not fully capture.

The challenge extends beyond faster change or larger volumes. Results in controlled tests may not represent behaviour in real operations. As systems interact and delegated permissions change, consequences can be difficult to trace, reproduce and measure. Understanding of some emerging risks—and the effectiveness of safeguards—is still developing.

Our perspective is that these conditions require us to reconsider how assurance is established and maintained. What evidence shows that systems are reliable, controls work and delegated tasks stay within their limits? How do we build our own justified confidence while making appropriate use of others’ assessments and independent challenge?

This survey is for business and technology owners, leaders, advisers and assurance professionals. It examines both the assurance you receive and what you do to establish a defensible basis for reliance. Your own checks complement independent challenge; they do not replace it.

Answer from the past 12 months, using one organisation and area of responsibility. Current approaches may remain effective in your context. Where AI effects have not been assessed, “not sure” is a useful answer.

All questions are optional. “Not sure” and “Not applicable” are valid answers. The opening states our working thesis; your answers should reflect your experience, including where that thesis does not hold.

Paper completion: mark one option unless instructed otherwise. For rating questions, mark one response per statement.