On a recent episode of This Week in AI , we discussed that the number of nonhumans on the internet is greater than humans, 144 to 1. It’s an estimate, and the order of magnitude is more important than the number itself. What matters is what a ratio anywhere in that range does to a security architecture. For more than two decades, systems were engineered around two foundational assumptions: either verifying a human user exercising discretionary judgment or validating the account executing hardcoded predictable logic. Autonomous AI agents shatter both sides of this legacy identity taxonomy because they possess broad, human-like operational scope across multiple internal and external applications while remaining inherently nondeterministic and probabilistic in their reasoning.

Today, because the agent is interacting on behalf of a human through a genuine browser with standard extensions, its technical fingerprint looks identical to a normal human user’s device. Agents call tools, observe results, and revise.…