Yubico has announced a strategic collaboration with IBM and Auth0 aimed at helping organisations securely deploy agentic AI systems at scale, introducing a new human‑verified authorisation model for high-risk automated actions. The partnership addresses a growing concern: as AI agents increasingly operate as autonomous “digital workers,” enterprises face heightened risks around governance, accountability and identity assurance.
With AI systems now capable of executing financial transactions, deploying production code, and accessing sensitive corporate data, the need for robust oversight has surged. Industry data indicates that 87% of organisations report increased risk from AI vulnerabilities, while 77% lack foundational AI security practices a gap the three companies aim to close.
The joint solution introduces a Human‑in‑the‑Loop (HITL) architecture that allows AI agents to function autonomously for routine tasks but requires cryptographically verified human approval for sensitive or high-impact actions. This approach ensures organisations can track not only what an AI system did, but precisely who authorised it.
“AI agents are becoming digital workers, but enterprises can’t afford to sacrifice trust, governance or human oversight.”
— Sheryl Chamberlain, Vice President, Yubico
At the heart of the model is Yubico’s hardware-backed authentication, powered by the YubiKey. When an AI‑initiated high-risk action occurs, Auth0 triggers a secure approval request, while IBM’s AI agents manage workflow orchestration. The designated human operator must physically authenticate using a YubiKey, confirming identity and providing non-repudiable proof of approval.
The collaboration supports key enterprise use cases including high‑value financial authorisations, production code deployment, data access, procurement approval, and security response actions where system availability may be affected.
Sheryl Chamberlain, Vice President at Yubico, said organisations are reaching a turning point as AI moves into the operational core. “AI agents are quickly becoming part of the enterprise workforce,” she said. “This partnership ensures the speed of automation doesn’t come at the cost of trust or governance.”
By uniting IBM’s WatsonX AI orchestration, Auth0’s identity workflows, and Yubico’s hardware root of trust, the companies aim to build a secure foundation for the emerging era of autonomous enterprise operations one where AI acts at machine speed, but humans remain firmly in control of critical decisions.