Jonathan Awad warns 90% will be exploited by AI agents as Baselayer raises $35M Series A

The problem: fraud systems were built for humans

Awad describes a scenario where an autonomous agent lands on Gap’s website and enters credit-card details on a user’s behalf. The core question: how does the merchant know the action was authorized by the cardholder, not by an attacker who stole the details? His argument: every existing fraud-prevention layer on the web was designed around human behavior — typing cadence, geolocation, device fingerprinting — and none of it knows how to handle a software entity acting for a legitimate person.

A field example: the Muse address leak

To illustrate the risk, Awad cites a Threads post describing how a tool called Muse exposed a Facebook Marketplace seller’s home address to potential buyers, who then showed up at the seller’s door. He calls the incident dangerous and creepy, and notes the harm would have been a thousand times worse had the victim been a woman. The example is meant to show that agents can leak sensitive personal data without any malicious intent, simply because there is no control layer governing what they share.

The solution: Agentic Identity

Baselayer proposes an infrastructure it calls Agentic Identity, built from three pieces. First, KYA — Know Your Agent — a mechanism that lets the receiving party identify which agent arrived and which human authorized its action. Second, Counterparty Verification, designed to let the agent itself confirm it is talking to a legitimate party and not falling victim to prompt injection or site-side fraud. Third, a Trusted Registry, a central registration layer intended to bind agent identities to their permissions.

What the announcement leaves out

The release does not name the investors in the round, disclose the product’s stage — whether there are paying customers or only a prototype — or specify which standards or protocols Baselayer plans to adopt so that such a registry could become an industry standard. Without details on architecture, trust model, or compatibility with existing payment rails, it is hard to assess whether KYA can succeed where previous digital-identity efforts have failed. For now it is a statement of intent backed by capital, not a demonstration of capability.