Sharing some learnings from my side. Felix set up a call with Sebastian Bode (LinkedIn), Director of Engineering at the Cardano Foundation.
Sebastian is currently building a Cosmos<>Cardano bridge and we discussed the bridge topics and I clarified some differences between IBC & XCM.
The most conservative/developed approach for the Cardano->Polkadot direction seems to be Mithril. Mithril(Paper, Github) is a research project of the Cardano foundation, which provides stake-based threshold multisignatures. A subset of Cardano validators opt-in to provide those signatures. In my simple monkey understanding, this is a light client(-ish?) protocol. Cardano transaction finality is achieved after one day. If I understand correctly, Mithril has shorter times to create state proofs, so this means that their light client protocol runs ahead of finality.
For the Polkadot->Cardano direction, I believe the most straightforward approach would be to verify Beefy proofs on a contract on Cardano L1.