Existing Machine Learning and AI algorithms can potentially be readapted to tackle similar issues in web3, leveraging on-chain data instead of web2 inputs.
Here we were able to develop a showcase Proprietary algorithm which was battle hardened in web2 anti-sybil use cases, and readapted for web3 use cases leverage account-level on-chain data provided by Hemera Network.
For more details of the approach, see the blog post here: