Identity at the Center #398 - Solving the AI Identity Challenge with Martin Kuppinger
Feb 2, 2026
Martin Kuppinger, principal analyst at KuppingerCole and identity expert, discusses AI agents and the growing disillusionment with AI in regulated industries. He explores why agents differ from traditional workload identities. Short takes cover governance and shadow agents, the identity fabric as a capability-based approach, and using telemetry to cut over-entitlements.
AI Snips
Chapters
Transcript
Episode notes
AI Disillusionment And Complexity
- Organizations face growing AI disillusionment because promised results haven't matched reality and control is lacking in regulated industries.
- Martin Kuppinger calls this the most intellectually challenging evolution in 35+ years of identity work due to complexity and lack of control.
Agents Are Not Just Machines
- AI agents differ from traditional workload identities by having far higher autonomy and forming chains of delegated actions.
- This creates many more authorization points and complex human-agent relationships than previous identity models.
Provide Agent Identities As A Service
- Deliver identity and access as a centrally managed service rather than letting everyone set agent permissions ad hoc.
- Register and control agents so data owners retain authority and decentralized agent sprawl is prevented.