More than half of surveyed business decision-makers say AI systems at their organisations can act without routine review, while 87% say their organisation needs better ways to approve, monitor and control AI use. The figures point to a widening governance challenge as AI moves from answering questions to carrying out work.
Watch Desk analysis
What happened
Help Net Security reports on a Vanson Bourne survey of 3,600 business decision-makers at companies with more than US$500 million in annual revenue, across seven sectors and several regions. Respondents reported that conversational AI and virtual agents were most likely to act unsupervised, at 59%; standalone generative tools came in at 54%. These are managers’ reports about their own teams, not an independent audit of deployed systems.
The survey also found that 46% of organisations that had deployed or scaled AI quickly reported security, privacy, regulatory or compliance problems. A quarter reported losing visibility or control of AI systems or their permissions. Separately, 87% said their organisation needed a better approach to approving, monitoring and controlling AI.
Why it matters
AI systems acting within defined rules can still create real operational consequences. If staff cannot see what an agent can access, who approved it or when it acted, “human oversight” risks becoming a reassuring label rather than a working control.
The results are self-reported and cover large companies, so they are not a census of AI use everywhere. Even so, the combination of reported autonomy and demand for better controls makes this more than another survey about enthusiasm for AI.
Our read
The striking gap is not that businesses want AI to move faster. It is that many say their systems already act with limited review while their controls need work. Before adding another agent to the org chart, organisations should be able to explain its permissions, monitoring and route to human intervention. Less glamorous than a launch video, rather more useful on Monday morning.
What to watch
- Whether companies disclose clearer rules for approving and monitoring AI agents.
- Whether future surveys distinguish reported autonomy from independently assessed deployment.
- Whether security and governance investment catches up with the pace of adoption. Activity teaser: More than half of surveyed business decision-makers say AI at their organisations can act without routine review, while 87% say their controls need to improve. The findings, reported by Help Net Security from a Vanson Bourne survey, cover large companies across seven sectors. They raise a practical question: when an AI agent acts, can its organisation see what it did and who is accountable? The full story looks at the figures and their limits.
Discussion spark: Should companies be allowed to deploy AI agents that act without routine review if they cannot yet show they can reliably monitor and control them?
Sources and evidence
- Source update (6 November 2026, 06:00 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.