Security Thoughts on Untrusted Nodes

One of the biggest challenges we will face relative to executing compute over data will be how to execute over encrypted/secure data.

We do not have a current plan of attack other than our current plan of:

1. Using public data (only) to start
2. Encouraging the use of subnets on trusted nodes.

Projects

However, there is quite a bit of thinking in the context of MPC (multi-party compute). Below is a partial list of folks who have already thought about it (in alpha order):

• ABY^3 - a semi-honest implementation of ABY 3 and Fast Database Joins for Secret Shared Data. https://github.com/ladnir/aby3
• Awesome Homomorphic Encryption (HE) - list of resources around HE https://github.com/jonaschn/awesome-he
• Batch DualEx - https://github.com/osu-crypto/batchDualEx
• Crypten - Privacy Preserving Machine Learning built on PyTorch https://github.com/facebookresearch/CrypTen
• EMP - Efficient, multi-party computation toolkit - https://github.com/emp-toolkit/emp-tool
• Honeybadger MPC - HoneyBadgerMPC is a robust MPC-based confidentiality layer for blockchains. https://github.com/initc3/HoneyBadgerMPC
• JIFF - Secure Javascript for multiparty compute - https://github.com/multiparty/jiff/
• MOTION - Mixed protocol multi party compute - https://github.com/encryptogroup/MOTION
• MPyC - multiparty compute in python - https://www.win.tue.nl/~berry/mpyc/
• Multi-protocol SPDZ - benchmarking for MPC - https://github.com/data61/MP-SPDZ
• Nucypher - enables proxy re-encryption without access to the private keys and/or plaintext https://www.nucypher.com/
• Obliv-C - Data oblivious computation with a gcc wrapper - https://oblivc.org/
• Rosetta - Privacy-preserving framework based on TF - https://github.com/LatticeX-Foundation/Rosetta/
• Swanky - Rust privacy preserving framework https://github.com/GaloisInc/swanky
• TF Encrypted - a framework for encrypted machine learning in TensorFlow https://github.com/tf-encrypted/tf-encrypted
• TF Federated Learning - https://ai.googleblog.com/2022/02/federated-learning-with-formal.html

Papers

• Refereed Delegation of Computation - https://www.cs.tau.ac.il/~canetti/CRR11.pdf
• Incentivizing Outsourced Computation - http://netecon.seas.harvard.edu/NetEcon08/Papers/Belenkiy_08.pdf
• Federated Learning Paradigm (2021-06): https://arxiv.org/pdf/2009.09338.pdf
• Integration of blockchain and federated learning for Internet of Things: Recent advances and future challenges (2021-09): https://www.sciencedirect.com/science/article/pii/S0167404821001796
• Comprehensive survey (2021-10): https://arxiv.org/abs/2110.02182
• Blockchain for federated learning toward secure distributed machine learning systems: a systemic survey - https://link.springer.com/content/pdf/10.1007/s00500-021-06496-5.pdf
• Blockchain-based federated learning methodologies in smart environments (2021-11) https://www.ncbi.nlm.nih.gov/labs/pmc/articles/PMC8561346/
• PRACTICAL IMPLEMENTATION OF BLOCKCHAIN ENABLED FEDERATED LEARNING (2020-09) https://udspace.udel.edu/bitstream/handle/19716/27516/Li%2C Wenxuan (Owen).pdf?sequence=1&isAllowed=y

Guidance

IPFS's guidance on improving privacy/security - https://docs.ipfs.io/concepts/privacy-and-encryption/#enhancing-your-privacy