Privacy matters: PETs for AI and a map of encrypted compute protocols
Thread297 words
Argues privacy-preserving technologies (MPC, ZK, FHE) are needed for AI on sensitive data and as AI agents dominate onchain activity. Lists encrypted-compute protocols for AI and separate private DeFi infrastructure.
Key claims
AI agents will dominate onchain activity
PETs unlock AI on financial and health data
Privacy matters. 🥷
Even more so in the age of (personalized) AI, where fine-tuned AI/ML models will increasingly work with sensitive, high-value data and where AI agents will dominate on-chain activity.
While advanced and potentially specialized AI models bear great potential to automate and optimize many aspects of our lives, ensuring privacy and maintaining compliance with data protection regulations globally, are no easy tasks when working with sensitive (financial, health, other personal) data.
Privacy-Preserving Technologies (PETs) such as MPC, ZK or FHE can (in various fields & forms of application), be a solution to these challenges, and can hence unlock the full potential of AI in the context of high-value data related use cases.
Recommend you to look into protocols that enable encrypted compute and private apps leveraging these above-mentioned technologies. Definitely a trend you can still catch early.
Some interesting protocols (among others), include:
➤ @nillion (blind compute network, leveraging various PETs)
➤ @zama (fully homomorphic encryption for blockchain & AI)
➤ @fhenix (FHE-powered L2 & Coprocessor)
➤ @OasisProtocol (confidential EVM)
➤ @inconetwork (confidentiality layer powered by FHE)
➤ @aztecnetwork (programmable privacy ZK L2 on Ethereum)
➤ @arcium (private compute for a hyper-encrypted internet)
➤ +++
While not directly related to AI but nevertheless interesting, there is also quite a few projects building infrastructure to enable private DeFi:
➤ @RAILGUN_Project (privacy for EVM assets & DeFi)
➤ @namada (private transactions & shielded DeFi)
➤ @penumbrazone (private interchain DEX)
➤ @DarkFiSquad (L1 for anonymous engineering)
➤ +++
Feel free to mention other interesting ones in the comments!
Still not sure why privacy even matters?
The below @TEDTalks remains the best explainer why you should care. ↓
will drop short explainers on FHE, MPC, TEEs later today, stay tuned! 👀
Linking back to these definitions here. ↓
x.com/zerokn0wledge_/status/1849828471624642726
1 image on X
What came next · since 25 Oct 2024
Since then
Zero has written 6 more pieces on Nillion since this one.