[jellyksong]

The article claims that oblivious query, set intersection, and machine learning on private data are not possible without FHE. However, aren't they all possible either with secure MPC or hardware based enclaves e.g. AMD SEV?

[Taek]

I would argue that secure enclaves do not exist in practice. You have to assume a physical device where the private key cannot be extracted and the operation cannot be observed. The threat model is much weaker than for FHE, and imo not really useful for operating in large sets of private data.

MPC introduces a trust assumption. MPC is only private if some threshold of the multiple parties are honest and destroy their secrets. Though often this threshold is just one, efficient MPC often only has 3 participants total.

FHE gives you a better security model than either of these, as it neither relies on physics for safety nor needs to assume any level of honesty from the person running the computation.

[Ar-Curunir]

In many MPC schemes you only need to trust yourself; as long as you keep your stuff private, nobody can learn the MPC secrets

[nullc]

FHE is also not obfuscation -- you can't just decrypt the output. Makes it a lot less useful than people usually imagine.

[bollu]

How does secure multiparty give access to obliviousness?

[jellyksong]

I was thinking MPC might work for set intersection. For oblivious query, can't you do it by sending the query directly into the remote trusted execution environment, encrypted with the TEE's public key?