Originally reported by Foresight News on November 13, 2026, at 23:34 (UTC+8). Circle Research has proposed Anduin, an experimental system for ranking AI-agent service providers using on-chain records of who paid whom. The aim is to make provider rankings harder to manipulate with fake identities, or Sybil attacks.
The research says existing ratings and usage metrics are easy to falsify: an attacker could raise a fake provider’s rating to the market median for about $0.0027. Under Anduin’s proposed approach, moving that provider up a particular AI agent’s ranking would require real payments to independent providers the agent already uses. Moving funds between an attacker’s own wallets would not earn reputation.
Circle tested Anduin against 66.1 million x402 payments totaling $6.1 million across 16 networks, from March 23 to August 26, 2026. Its recommendation accuracy was close to that of collaborative filtering, the strongest conventional method in the comparison. In the attack scenarios the research tested, manipulation cost about 23 times as much as with collaborative filtering.
Circle says Anduin remains experimental and would not require changes to the x402 protocol or ERC standards, or the issuance of a new token. The reported results are research findings, not a deployed protection for users choosing a service provider.
Source: foresightnews.pro (originally published 2026-11-13 23:34 UTC+8)
