Bloom filters test for membership in a set. A tutorial describes this better.
Compute several hashes across each of the members of the set. Store those hashes in a bit vector. You don’t need a cryptographically strong hash, but you do want it to be fast.
To test for set membership, compute a hash of the candidate item. Look up that hash in the bit vector. If the bits are not set, you know that the candidate is not in the set.
There is a chance of false positives if you have a hash collision, which you can reduce by making the bit vector longer.