Chain Tech Catcher

What the timeline buried, caught once.

Circle Research’s Anduin Ranks AI Agent Services by Payment Relationships

Circle Research has proposed Anduin, an experimental reputation system for AI Agent service providers. Rather than relying on ratings or usage counts, it ranks providers using the on-chain record of who paid whom. The aim is to make it harder for an attacker to boost a fake provider’s standing with fabricated activity.

According to the research, an attacker could push a fake provider’s rating to the market median for about $0.0027 under existing metrics. With Anduin, moving that provider up in a target AI Agent’s ranking would require real payments to independent providers the agent already uses. Moving funds between the attacker’s own wallets would not build reputation.

Circle Research’s Anduin Ranks AI Agent Services by Payment Relationships (1)

Circle Research tested the approach on 66.1 million x402 payments totaling $6.1 million across 16 networks, dated March 23–August 26, 2026. It reported recommendation accuracy close to the best-performing traditional method, collaborative filtering. In the attack scenarios it tested, manipulation cost about 23 times as much as with that method. Those results describe a test, not a deployed system: Circle said Anduin remained experimental and would require no changes to the x402 protocol or ERC standards, and no new token.

Circle Research’s Anduin Ranks AI Agent Services by Payment Relationships (2)

Source: foresightnews.pro (originally published 2026-11-13 23:34 UTC+8)