Delinea’s 2026 Report Finds AI Identity Security Confidence Outpaces Governance
Report discovered via Awesome Annual Security Reports.
Summary
Delinea’s survey of 2,001 IT decision-makers found 87% considered their identity security ready for AI automation, yet 46% reported deficient AI identity governance. Shadow AI and standing privileges compound oversight gaps.
Key points
- Delinea commissioned Censuswide to survey 2,001 IT decision-makers actively using or piloting AI across seven countries.
- While 87% reported readiness for AI-driven automation at scale, 46% acknowledged deficient identity governance around AI systems.
- Some 90% reported identity visibility gaps; 82% were very confident in non-human identity discovery, but only 30% validated NHI and AI usage in real time.
- Among respondents, 53% regularly encountered unsanctioned AI tools or agents accessing company systems or data, while 28% reported real-time shadow AI detection.
- Nine in ten reported pressure to loosen access controls for AI automation; fewer than one in three said security requirements were consistently enforced when they conflicted with business speed.
- Long-lived credentials were the most common access method for 35%, compared with 17% using just-in-time authorization and 8% using ephemeral access; 80% could not always explain privileged NHI actions.
- The report recommends continuous identity visibility, just-in-time and ephemeral access, microsegmentation, isolated experimentation and limits on agent autonomy.
Article Details
- Publisher
- Delinea; Censuswide conducted the survey
- Report Period
- 2026
- Scope
- Survey of IT decision-makers actively using or piloting AI, covering identity security, AI-related identities, and governance.
- Sample Size
- 2,001 IT decision-makers across the UK, US, Germany, France, Australia, Singapore, and India.
- Key Statistics
- 90% of respondents said their organization has at least some identity visibility gap.
- 51% reported persistent identity-discovery gaps in AI-related environments.
- 46% said their identity governance is deficient around AI systems.
- 53% regularly encounter unsanctioned AI tools or agents accessing company systems or data.
- 73% agreed that standing access for NHIs and AI agents increases risk.
- Recommendations
- Improve visibility into identities, privileges, and activity, including signs of shadow AI and unexpected privilege requests.
- Use machine-speed security controls to address the scale and speed of AI-agent activity.
- Move toward just-in-time and ephemeral access instead of static, long-lived credentials and standing privileges.
- Apply least-permissive access, microsegmentation, and limits on agent autonomy to constrain actions and potential impact.
- Provide isolated environments with synthetic or public data for AI experimentation.
MITRE ATT&CK
People
Chris HughesQuoted expert at Resilient Cyber; discussed AI-agent permissions, identity governance, and security controls.Gal DiskinIdentified as VP, Identity Threat Product & Research and head of Delinea Labs; discussed identity supply-chain attacks.Gerald AugerQuoted expert and head of Simply Cyber; discussed identity risk and visibility gaps.Kayla WilliamsQuoted vCISO and SANS Institute’s 2024 CISO of the year; discussed identity visibility and standing access risks.
Vendors
Products
Countries
Australiawho are actively using or piloting AI in their environments across the UK, US, Germany, France, Australia, Singapore, and India.Francedecision-makers who are actively using or piloting AI in their environments across the UK, US, Germany, France, Australia, Singapore, and India.Germany2,001 IT decision-makers who are actively using or piloting AI in their environments across the UK, US, Germany, France, Australia, Singapore, and India.Indiausing or piloting AI in their environments across the UK, US, Germany, France, Australia, Singapore, and India.Singaporeare actively using or piloting AI in their environments across the UK, US, Germany, France, Australia, Singapore, and India.United Kingdomsurvey of 2,001 IT decision-makers who are actively using or piloting AI in their environments across the UK, US, Germany, France, Australia, Singapore, and India.United Statesof 2,001 IT decision-makers who are actively using or piloting AI in their environments across the UK, US, Germany, France, Australia, Singapore, and India.