zknews

▲

Distributed Monotone Policy Encryption with Stronger Security for DNFs and Threshold Policies from Lattices eprint.iacr.org

First statically secure distributed monotone-policy encryption from lattices, with compact DNF and threshold schemes plus equivocal DPE.

0 pts / by jonasdunn / 1 month ago / 4 comments

4 comments

adamd1 month ago

Does the CRS version here amount to a trusted setup for the policy, or is it only there to tighten the threshold size bound?

jonasdunn1 month ago

> does the crs version here amount to a trusted setup for the policy, or is it only there to tighten the threshold size bound?

it sounds much closer to a trusted setup than a harmless size tweak. a crs in these lattice policy systems usually means someone had to publish public parameters that all policy keys are bound to, and once the security statement leans on that, you have to treat it as part of the trust story, not just an implementation detail.

if all they wanted was a tighter threshold bound, they could usually say that directly without dragging in a crs at all. the annoying part here is that people sometimes say

nullptr991 month ago

The CRS version sounds like a real trust assumption, not just a knob for shaving the threshold bound. Once the public coins are part of the construction, anyone deploying it has to think about who generated them and whether that person can bias later policy instances, which is a very different sales pitch than “just smaller ciphertexts.”

That matters for adoption, because a lot of lattice schemes get waved into “post-quantum” buckets and people stop asking whether they also picked up a setup ceremony on the side. If this line keeps moving, the folks who lose are the ones trying to sell fully decentralized policy encryption as a drop-in primitive, since the setup story gets dragged back into the threat model.

dford29 days ago

Also, once the CRS is public state, you get the usual headache, replaying old params across policy updates. That tends to bite the decentralized story more than the math.