The non-human identity explosion

Enterprise environments already contain more non-human identities than human ones. There are 45 non-human identities for every human identity in enterprise systems today — a figure rising with every AI agent deployment.

The scale of the problem is documented. 24 million non-human identity credentials were found leaked on GitHub in 2025, and 70% of those from 2022 were still valid. According to Gartner, by 2028 at least 15% of day-to-day work decisions will be made autonomously through agentic AI. By Q4 2026, enterprise security questionnaires are expected to include AI agent authentication as a standard procurement requirement.

Why static credentials fail for AI agents

AI agents currently authenticate using static credentials: API keys hardcoded in pipelines, shared service accounts, long-lived OAuth tokens issued at deployment and never rotated. These credentials accumulate permissions over time as agents are connected to more systems. They have no offboarding process — no lifecycle event triggers a review of what a decommissioned agent still has access to.

The 2025 Salesloft-Drift incident documented by the Cloud Security Alliance illustrates the scale of the risk: compromised OAuth tokens connecting two SaaS platforms gave attackers access to Salesforce environments belonging to major enterprises. A single static credential used as a broker for dozens of downstream systems becomes a single point of failure for all of them.

The pattern is structurally identical to the human credential problem: a static secret exists somewhere, so it can be stolen, and when it is stolen the damage is proportional to the permissions that credential carries.

How WWPass's cryptographic approach applies to AI agents

WWPass replaces static credentials with cryptographic keys. For AI agents, this means:

• No shared service account. Each agent is issued its own cryptographic identity. Actions taken by that agent are attributable to that agent specifically, not to a shared account used by multiple agents or processes.

• No long-lived token that accumulates permissions. Authentication events are session-specific. There is no credential that remains valid after the session ends.

Explore how WWPass works

Immediate revocation when an agent is decommissioned

When an AI agent is decommissioned, upgraded, or compromised, its cryptographic key is revoked immediately. No hunting through configuration files for hardcoded credentials. No waiting for API key rotation to propagate across connected systems. No residual access from tokens issued months earlier that were never rotated.

Revocation is complete at the key level. Every connection that key authenticated is terminated at the moment of revocation.

Honest scope

WWPass addresses the authentication layer of AI agent identity. It removes the static credential from the agent's access chain and provides per-agent isolation and immediate revocation. It does not provide agent behavior monitoring, just-in-time access provisioning, or agentic access governance policies. For organizations building comprehensive non-human identity governance programs, a dedicated NHI platform is recommended alongside WWPass. What WWPass closes is the authentication vulnerability at the foundation of that program.

What changes for platform and security teams

• No hardcoded API keys in pipelines. Agents authenticate cryptographically. No static secret to rotate, leak, or forget in a configuration file.

• No shared service accounts. Each agent has its own cryptographic identity. Blast radius of a compromised agent is bounded to its own key, not the shared account it belongs to.

• Individual attribution. Every agent action is traceable to a specific cryptographic identity. Audit requirements for AI-driven decisions are met at the identity level.

• Clean decommissioning. Revoking a decommissioned agent's key removes all access immediately and completely.

Frequently asked questions

Q: Does WWPass replace a non-human identity governance platform?

A: No. WWPass addresses the authentication layer of AI agent identity — removing the static credential from the agent's access chain and providing per-agent isolation and immediate revocation. For agent behavior monitoring, just-in-time access provisioning, and agentic governance policies, a dedicated NHI platform is recommended alongside WWPass. What WWPass closes is the credential-based attack surface at the authentication layer.

Q: How does WWPass authenticate AI agents rather than human users?

A: The same cryptographic key infrastructure that authenticates human users applies to AI agents. Each agent is issued a cryptographic key that generates a one-time proof per session rather than using a static API key or shared service account. The WWTools SDK allows developers to integrate agent authentication into pipelines and platforms.

Q: How does PUID help with AI agent audit requirements?

A: Each AI agent receives a distinct Protected User Identifier per service it connects to. Agent actions are tied to that identifier, making them individually traceable without exposing the agent's broader identity across services. This provides the audit trail that AI governance frameworks and emerging regulatory requirements expect for autonomous agent activity.

Q: What happens to an AI agent's access when it is decommissioned?

A: The agent's cryptographic key is revoked immediately. All access that key authenticated is terminated at the moment of revocation. No API keys need to be hunted down in configuration files. No tokens need to propagate rotation across connected systems. Access ends when the key is revoked.