A thread on Privacy-Preserving Machine Learning with Fully Homomorphic Encryption (say that 3 times fast) 1/ Most LLMs (Large Language Models) today are trained primarily on publicly available data. This can limit their applicability in domains with strict data privacy

Play the attention game, don’t get played in the attention game

My daily hustles….

Since a lot of people are interested in fully homomorphic encryption (FHE) recently, re-upping my article from 2020 that goes into how the math works in depth: vitalik.eth.limo/general/2020/0…

Noticed some AI x Crypto projects projects are deliberately confusing the concepts of the Alan Turing Development Award and the Turing Award. 🥱

Heard from Will Robinson | Alliance ✈️ DevCon that 200+ L2s, all of them EVM compatible, are launching this year. Only one Cairo-based Starknet 🐺🐱 L2. That makes you more...

Due to information loss in analogizing

we need fhe

Agree and disagree. Agree De-Disney didn’t work out. Disagree “no creation but curation”. Counterexample: TikTok. Every minute, new IPs are created, consumed and forgotten in a decentralized manner. Oversupply and unlimited supply is not a problem bcz human memory is FIFO too.

🌐 Join us at the Symposium to hear Wei Dai from 1kx lead a discussion on how shared security unlocks Better Secure Computation, with panelists Jay Prakash from Silence Laboratories S El Damaty from Holonym Brz from Sight AI Remi Gai from Inco

This would be great

😂

A prev Google friend asked me a question: "why working on improving privacy, isn't it slowing down product shipping & growth?" I'll probably write an article to explain that safe data-sharing actually helps growth simply bcz the velocity of data will be faster in data economy.

1st day EDCON De University of Ethereum

Elon Musk Blockchain!

He has demonstrated he can learn any area and mostly do better than experts who mostly figure out what can’t be done. Learning rapidly is always more important than knowing, especially knowing a previous version of the world which experts are experts at.

Trust math

🚨 The winning team of the Tik Tok TechJam Hackathon built a privacy-preserving search engine using FHE and Zama's Concrete ML library. This is big 👏 tiktoktechjam2024.devpost.com/project-gallery

gm. We're super exited to announce a new way to read & write to web3: 🌈 simple. free. ⚡️ up to 1 million requests/day ☁️ global scale 🔜 multi-chain give it a try! cloud.google.com/application/we… Ethereum Singapore TOKEN2049