09/02/2026
94% of enterprises are confident their AI agents don't have too much access. Only 33% have actually verified it.
New research surveying 202 enterprise security leaders found that gap isn't just theoretical: 65% have already seen an AI agent act outside its intended scope, and 29% saw real damage, data exposure, financial loss, or worse. Most companies still need hours and manual steps to catch it.
Contact us if you'd like to review your security posture.
EMA survey finds 65% of enterprises have seen AI agents act beyond intended scope