Ping Identity Proposes Dual-Identity Framework for AI Agents

Suman Sharma, Head of PAM Engineering at Ping Identity, has proposed a dual-identity framework designed to secure AI agents, bots, and IoT devices. The model integrates non-human identity (NHI) with human identity to balance operational functionality with security and accountability.

Under this framework, NHI provides agents with unique credentials, such as cryptographic tokens or API keys, to ensure traceability and governance. Simultaneously, human identity allows agents to execute user-specific tasks, such as accessing payment information or travel APIs, using credentials like OAuth tokens. Sharma argues this approach is essential for personalization, particularly within highly regulated sectors like healthcare and finance.

Despite the benefits, the framework introduces risks including privilege escalation and identity sprawl. Sharma notes that these vulnerabilities require robust encryption and management systems to mitigate. The integration of these identities is presented as a critical step as AI evolves toward artificial general intelligence. This proposal aligns with broader industry trends, including a 2025 Accenture survey indicating that 78% of executives believe digital ecosystems must evolve to accommodate AI agents.