IPFS is a content-addressed peer-to-peer data network, which follows the paradigm of information centric networking. In IPFS, data is exchanged with the Bitswap protocol. For content discovery, Bitswap queries all neighbors for the content, leaking the interest to all neighbors. In our paper, we develop three privacy-enhanced protocols for content discovery, which reduce the interest leak from all neighbors to ideally one content provider. Our protocols use probabilistic data structures like Bloom filter and cryptographic approaches like Private Set Intersection. We implement our protocols as proof of concept and show how they can be integrated into the go implementation of Bitswap. Furthermore, we provide a measurement supported performance, load, and privacy evaluation of the three protocols, showing their feasibility trade-offs.