AI Adoption Forces Trade-Off Between Speed and Identity Security, Study Finds
AI adoption is forcing enterprises to trade security for speed — and identity controls are the first casualty, according to a new report from Delinea, a provider of identity security solutions for both human and AI agent identities.
A key finding in the 2026 Identity Security Report says 90% of organizations are forcing their security teams to loosen identity controls for AI.
In simpler terms, organizations are prioritizing speed over security in deploying AI tools, with leadership focused on faster adoption to drive productivity gains.
The major problem is that it leaves organizations heavily exposed to security vulnerabilities. Enterprises are fast-tracking AI initiatives, despite significant gaps in AI identity discovery, monitoring, and privilege control.
"The pressure to move fast on AI is real, but identity governance has not kept pace, which exposes enterprises to significant risk," said Delinia CEO Art Gilliland.
Over 2,000 IT decision-makers actively using or piloting AI were surveyed by Delinea. According to the report, 90% of respondents had at least one identity visibility gap, with the largest gap tied to machine and non-human identities (NHIs), including accounts used by AI agents.
"As AI agents multiply across enterprise environments, these identities often have the least oversight," Gilliland said. "The organizations that will succeed in the AI era will be the ones that enforce real-time, contextual access across every human, machine, and agentic AI identity."
Other findings from the report include:
- AI expansion is driving non-human identity risk: 42% of organizations said AI expansion has been one of the top factors increasing NHI risk in the past 12 months, far surpassing increased automation and CI/CD velocity (26%) and growth in cloud-native workloads (26%).
- Limited visibility into privileged AI actions: 80% of organizations said they are unable always to understand why an NHI performed a privileged action, highlighting major challenges with traceability and accountability for automated identities.
- Standing access remains the norm: 59% of organizations reported lacking viable alternatives to standing privileged access for NHIs and AI agents, increasing the risk that automated identities retain persistent permissions that could be exploited.
The result of all this is that traditional identity protections haven't kept up with AI and loosening identity controls has provided bad actors with an exponentially larger attack surface.
The report concluded that AI will continue to break traditional security models as companies allow their security controls to grow lax and more identities and access points appear.
"Clearly, organizations can't afford to slow down AI adoption," Delinea said. "But the study indicates that identity security must evolve alongside AI adoption."
The full report is available here on the Delinea site.