We’re excited to announce our partnership with
@zama_fhe
to expand
#FHE
capabilities in AI! 🤝
Zama is developing FHE solutions for blockchain and AI and is already being recognised for industry leading innovation.
Read below to understand what our partnership will entail
As use-cases for decentralized AI grow, so too does the urgent need to address verifiability, security and privacy.
Verified
#FHE
represents a critical step forward in securing AI-powered smart contracts, enabling them to execute more complex tasks while ensuring data integrity.
🟢 Verifiable FHE
Verifiable
#FHE
is essential to ensure integrity on blockchains.
SPARC, Structure-Preserved Arguments for Computing, is an efficient proving system for verifiable FHE.
This ongoing joint research effort is led by researchers from Sight AI, Zama, USC CS Theory
🟢 Sight AI Retro Campaign is Live
Link:
We are excited to announce the community
#retro
campaign.
Dive into the Sight AI
@zealy_io
page and start the journey with us.
Early adopters will get benefits
#zealy
Verifiable
#FHE
unlocks the door for millions of businesses and individuals to leverage the power of decentralized computation. 🗝️
Our partnership with
@zama_fhe
means deeper R&D into FHE; we're building towards a faster, more secure and accessible future for on-chain AI dApps.
Friendly reminder that our
@zealy_io
campaign is still live
Complete quests and WIN your share of $1000 $USDT 💸
Start here:
Complete tasks for rewards as the Sight AI community continues to grow!
#web3
#fhe
Kick off your week with some tech optimism!🚀
Here are 10 reasons to be bullish:
1. Pioneering verifiable FHE;
2. FHE Oracle for multi-chain support;
3. SPARC innovation for efficient proofs;
4. Secure on-chain AI;
5. Confidential real-world asset tokenization;
6.
Curious to know what's been researched on Fully Homomorphic Encryption out there?
Academic papers discuss it in terms of data privacy and security across multiple fields, as well as practical applications like cloud computing and privacy-preserving machine learning.
The
We're thrilled to be developing
@theSightAI
#FHE
Oracle, pushing the boundaries of decentralized
#AI
.
🔏 Secure inference with a two-tiered architecture
⛓️Lightweight prove-verify protocol based on SPARC and SNARG
🗝️Reliable networks with Proof of Computation
1/2
5/10
"That's great, but why do u need ur own blockchain?"
Here is why, anon.
AI inference is resource intensive, requires performant GPUs, high data throughput & low latency - all things that legacy blockchains are not designed for.
As an EVM compatible blockchain dedicated to
Across an inherently trustless landscape, verification and security of data, AI models & their outputs is of paramount importance if decentralized AI is to continue its growth trajectory.
vFHE is a solution promising to open the floodgates for innovation in deAI.
How so? 👇
Explore these innovations driving the future of
#blockchain
⛓️
From confidential
#DeFi
and blind auctions to secure on-chain
#AI
and private RWA tokenization💡
Dive in and tell us: which excites you most?🤔
Congratulations to our
@zealy_io
reward winners!🏆
Your dedication is helping shape the future of privacy-preserving
#AI
🫂
Together, we're building a more secure and innovative AI ecosystem🍻
Check the winners' transaction IDs 👇
#EthCC
Talks
#1
: Unveiling the Future of Web3 Infrastructure🚀
Excited to share EthCC insights with our community!
Let's start by covering one of our key goals for
#Web3
infrastructure:
🗣️: "We aim to create a shop displaying all options for developers and end-users, with
FHE innovation is accelerating quickly and we're excited to be at the forefront of developments in this important area of work!
Our joint efforts with
@Google
Research and
@USC
CS Theory utilising SNARG for Ring has highlighted lower computational requirements and faster proof
1/ FHE for blockchains have came a long way since Vitalik's article from almost 4 years ago!
All three areas list by Vitalik have progressed further:
1. Threshold FHE for Blockchains
2. Oblivious Message Retrieval
3. Privacy-preserving data marketplace
More details👇
1/6 Verifiable FHE (vFHE) represents the pinnacle of data verifiability & computational privacy for web3 AI. 🙌
But why?
Before that, we advise anyone new to FHE to check out this great introductory article by
@0xMustafa_eth
:
More on vFHE and web3 AI 👇
2/10
In the web3 AI paradigm, computational tasks are distributed across a network of nodes, making robust privacy preservation a vital necessity.
AI models are exposed to theft, reverse-engineering and adversarial attacks from malicious actors - SightAI proposes a solution.
Our
#FHE
Swiss Army Knife takes secure computation to a higher level🔐
It supports multiple FHE schemes like CKKS (for precise arithmetic on encrypted real numbers) and TFHE (efficient for Boolean and integer operations), empowering
#developers
to choose the perfect tool for
10/10
That's it. That's the tweet. 🫡
Expect updates on our development roadmap, research publications and much more in the coming weeks.
We're excited to talk with anyone who is interested in the future of verified FHE for web3 AI!
Join the SightAI Telegram community:
4/10
Honesty and integrity cannot be assumed in decentralized AI - Verification matters.
❌ ZK-SNARK provides data verifiability but challenges arise when integrating FHE & ZK-SNARK.
✅ SightAI aligns SNARG with FHE to create vFHE, a lightweight prove-verify protocol that
9/10
Our Background
The SightAI team includes Founding Engineers, Professors, PhDs, cryptography researchers and growth marketers from the likes of Google, YouTube, USC, Harvard and Berkley.
We're also advised and supported by recognized figures from Stanford, Tsinghua
8/10
Multi-Layer OPML
Classical OPML supports inference on a single node, verifying integrity of a whole model inference process, however in a "Collaborative Inference" scenario where multiple nodes' effort are put together, it is hard to figure out which node is to be punished
7/10
Collaborative Inference
The SightAI Inference Network comprises Subnets, each hosting a subset of model layers (typically Transformer Blocks). Collaborator nodes break down inference tasks to a sequence of subnets for execution.
SightAI facilitates equitable participation
6/10
SightAI enables secure, private web3 inference with a 3 tiered architecture.
AI dApps request off-chain computations on AI models, whose results are then relayed to on-chain smart contracts.
- Prioritizes data availability
- Improves efficiency & utilization of AI compute
3/10
Fully Homomorphic Encryption (
#FHE
) is a hot topic in web3.
FHEML enables privacy-preserving machine learning in AI, allowing data, models & predictions to remain encrypted throughout the ML lifecycle. 🕵️♂️
But whilst FHE ensures privacy, it does NOT ensure integrity of
vFHEML enables truly secure computation and represents the holiest of grails in the cryptography world.
We're unlocking a world of new use cases for web3 AI inference with the introduction of the Sight AI blockchain and Inference Network.
Learn more:
How do you catalyse the next wave of innovation in web3 AI?
- Better security
- More privacy
- Data verifiability
- Affordable computation
- Developer friendly tooling
- Collaborative environments
Learn how Sight AI achieves all of the above and more:
Ensuring verifiability of output in web3 AI is vital if the industry is going to achieve real adoption.
Sight AI vFHE:
- Eliminates the need for difficult zkSNARK integration with FHE
- Ensures public verifiability
- Speeds up proof generation
- Lowers computational requirement
We're excited to today announce our collaboration with
@babylon_chain
! 🤝
By integrating Babylon's technology, Sight AI can document transactions during computation tasks with Bitcoin's security, enhanced reliability and trustworthiness.
Learn more:
The recent incident involving
@OpenAI
and Scarlett Johansson highlights the urgent need for robust data privacy and encryption measures in the AI industry.
How can verifiable FHE help in the battle against malicious actors in AI? 👇
Verifiable
#FHE
🟢 Sight AI Zealy Winners
Link:
You can find yourself by your Zealy username or by your EVM wallet address linked to Zealy. Press Ctrl+C or Cmd+C to do so.
USDT token will be sent to your EVM wallet connected to zealy.
Thanks, everyone for
1/9 It seems that everyone and their dog is shouting about
#FHE
and how it's going to change the face of computation.
But what is FHE and why should you care about it?
Let us answer those questions in the following thread 🧵
🟢 The First Sight AI Zealy Campaign has finished
Results and the leaderboard will be announced next week
Thanks, everyone for participating in this sprint, and have a good weekend!
#sightai
#FHE
#zealy
Verifiable Fully Homomorphic Encryption (vFHE) allows computations on encrypted data but also includes mechanisms to verify that these computations were carried out correctly.
🧵 Here are 5 ways that vFHE will transform Web3 AI:
Advisor Spotlight 🔦
Bruce Pon, Founder
@oceanprotocol
“Imagine a Google search where the search has the intuition to suggest a logical next step and execute it for you.
#FHE
is at the forefront of using AI to make this a reality; I’m excited to support Brian and the team.”
Step 1: Get $ETH
#Ethereum
Sepolia test tokens 💰
• Visit the faucet page:
• Enter your wallet address
• Wait for 0.05 ETH to be credited (limit: once per 24 hours)
2/8
Time to dive back into this gem - the
@theSightAI
&
@babylon_chain
collab📝Here are some great insights worth revisiting👇
The article discusses:
1. Babylon's
#Bitcoin
timestamping for Sight AI's task verification;
2. FHE enabling
#AI
on encrypted data;
3. Babylon's Bitcoin
🟢 We've been a part of
@shared_security
Thanks for the amazing experience and for the opportunity to share our ideas and thoughts with other enthusiasts all over the space
#FHE
#web3
🌟 Big News!
@theSightAI
is teaming up with
@0xBountyBay
!
BountyBay is a revolutionary platform built on the Telegram and TON ecosystem, designed for social viral marketing. It is supported by a
#TON
grant and the accelerator program co-hosted by the TON Foundation and Hashkey
6/7 Backed by considerable study with teams at Google Research Lab and USC CS Theory Group, Sight AI use of SNARGs enables efficient publicly verifiable FHE, with no field simulation overheads, proving superior to various other approaches currently being taken to vFHE.
7/7 In the coming weeks expect more info into what lies under the hood of Sight AI:
- Sight Chain
- Collaborative Inference Network
- DA Layer
- Heterogenous OPML
- Sight Intent model
- Native dApps
Until then, join our Telegram community:
Have you ever imagined computing encrypted
#data
across multiple blockchains without compromising privacy?
Transparency and security are key in the
#crypto
space, so the ability to compute on encrypted data without revealing the underlying information is big🍻
Using
Introducing
#FHE
FOMO: An on-chain game leveraging FHE and the Sight Oracle.
Winner of each pool will get 5000 points reward🏆
Participants deposit $ETH to match a hidden target while FHE ensures verifiable fairness for every transaction.
The game is on ETH-Sepolia testnet (no
6/6 You can learn more about what we're building by checking out our Gitbook:
And we welcome everyone to join in the conversation by joining our Telegram community!
4/6 Our innovations in vFHE developed alongside
@google
Research and
@USC
CS Theory Group propose leveraging SNARG on rings to speed up proof generation for AI inference.
This method improves upon SNARK implementation considerably; goodbye Ring and Field structure mismatches!
Step 4: Select a Pool 🏊♂️
• Click "Select Pool"
• View pool info and filter by:
Launched: Open pools
Completed: Closed pools
Is Winner: Your winning pools
• Choose a "Launched" pool to play
6/8
Here are some examples:
✅ Compatible with various heterogeneous blockchain networks;
✅ Supports multiple FHE schemes like CKKS and TFHE;
✅ Ensures security and fairness for the decentralized web.
4/5
The
#Ethereum
Book identifies two main types of oracles:
1️⃣ Data oracles: Provide off-chain data to smart contracts;
2️⃣ Computation oracles: Perform complex computations off-chain.
By lying between
#data
and computational oracles,
@theSightAI
Oracle represents a hybrid approach
2. Privacy-Preserving Verifiability: Verify the RNG process immediately without disclosing it;
3. Improved fairness: Cannot be manipulated or predicted.
Secure and verifiable randomness for dApps is here🔏
At a recent
#blockchain
event, someone asked:
"Why does Sight offer Fully Homomorphic Encryption
#FHE
through an Oracle rather than setting up a dedicated FHE blockchain?"
It was a great question! This decision allows us to leverage existing projects, users, and TVL on
Great conversation with
@CrustNetwork
on how we’re transforming decentralized storage and computation!
If you missed the live session, catch the recap and get up to speed on all the exciting updates. 🚀✨
Missed the Crust and
@theSightAI
Twitter Space? 🤔 Catch up with our detailed discussion recap and learn more about the collaboration between the two. Don’t miss out on the insights and updates! ✨
The crypto world needs dApps. They enable diverse applications extending
#blockchain
's utility beyond financial transactions.
However, did you know that dApps face a big challenge called fair "Random Number Generation (RNG)" ?
Let us explain better 👇
1/4
SNARGs are cryptographic methods that allow one party to demonstrate to another that a specific computation was executed correctly.
This is essential for verifying the integrity of decentralized AI computations.
Learn more about SNARGs and Sight AI:
1/9 It seems that everyone and their dog is shouting about
#FHE
and how it's going to change the face of computation.
But what is FHE and why should you care about it?
Let us answer those questions in the following thread 🧵
5/ Adoption of deAI in Sensitive Industries 🛡️
Industries such as healthcare, finance & government often face barriers to adopting deAI due to security concerns.
Verifiable
#FHE
mitigates these concerns, ensuring data can be used for computations without being decrypted.
Whenever you play
#blockchain
games, do you worry about keeping your strategies secret?
Our team is working on some innovative privacy-enhancing technologies:
1️⃣ Zero-Knowledge Proofs (ZK): Prove your moves are legit without revealing your strategy;
2️⃣ Fully Homomorphic
SPARC (Structure Preserved Arguments of Computation) is revolutionizing proof systems.
It leverages preserving computational structures to enable efficient proofs for
#FHE
where traditional SNARGs struggle.
But how does this translate to real-world
#blockchain
use?
1/2
Ever feel like understanding
#crypto
tech is like solving a puzzle?🧩
Once the puzzle pieces click into place, it all makes sense. That's
@theSightAI
for you - making complex crypto privacy simple👏
We're bringing secure on-chain
#AI
and confidential
#DeFi
to life, turning
The future of decentralized AI networks depends on technologies like verified
#FHE
.
Protecting sensitive data while leveraging value for intelligent decision-making is key to onboard businesses looking to adopt deAI.
Learn more about Sight AI and vFHE:
🌟 Exciting Partnership News!
@theSightAI
is teaming up with
@CrustNetwork
!
Follow & join TG➡️ and witness this groundbreaking collaboration.
Explore Crust Network➡️
Crust Network offers a decentralized storage network for the Web3
4/7 However
#FHE
itself guarantees zero knowledge; the verifier involved never requires proof of knowledge and this ensures public verifiability.
We have leveraged this to design an entirely new structure that disregards zero knowledge considerations from the start.
SightAI Oracle is a solution that combines secure
#data
handling and powerful computation⚡
A Computational Oracle that lets you delegate computations on encrypted data using Fully Homomorphic Encryption
#FHE
🔏
What makes Sight Oracle stand out?
3/5
Verifiable
#FHE
dramatically enhances the security, compliance, and functionality of AI dApps and systems in Web3, all of which are crucial to adoption of decentralized AI.
Discover how Sight AI is pioneering vFHE for decentralized AI inference:
Great to see
@hosseeb
working to demystify approaches to deAI inference (and why it all even matters) for a wider audience. 👏
DeAI is complex and evolving fast but the core challenge of verifiability is something that, when solved, will catalyse a whole new wave of innovation!
My talk @ Consensus on Decentralized AI Inference!
If you want a brief overview of crypto x AI and why it's such a thorny intersection, give it a watch (20 min) 👇
Oracles are vital in
#crypto
, connecting on-chain and off-chain worlds.
@theSightAI
Oracle takes this further offering secure, private, AND verifiable connections. How?
By leveraging cutting-edge tech: Fully Homomorphic Encryption (FHE). Let's explore...
1/2
3/ Improved Compliance with Data Privacy Regulations 🏛️
Regulations like GDPR & HIPAA require stringent data protection measures.
verifirable
#FHE
enables AI dApps to fully comply with these regulations, ensuring that personal or sensitive data isn't exposed during processing.
Let's talk about Oracles in the
#Ethereum
ecosystem🗣️
Smart contracts need Oracles to get external
#data
, as discussed in Ethereum Book here:
#Blockchain
1/5 🧵
As we continue to push the boundaries of what's possible in the
#Ethereum
ecosystem, innovations like these will be fundamental to shape the future of
#Web3
and
#DeFi
.
Join us for more:
Well, putting it simply:
1️⃣
#FHE
FOMO shows how
#blockchain
games stay fair and secure;
2️⃣ Sight Oracle uses Fully Homomorphic Encryption to keep everything under wraps until game over.
Are you curious about the smart contract code behind this? 👇
#coding
2/3
4/ Enabling Complex AI Applications 🧮
Verifiable
#FHE
facilitates complex AI applications on decentralized platforms, allowing multiple parties to collaboratively engage without compromising the confidentiality of their data, a valuable solution for many industries.
Step 3: Wallet Login 🔑
• Go to
• Click "Connect Wallet" in the top right
• Choose your preferred wallet
• Your wallet info will appear after connecting
5/8
3/7 The above approach makes it very challenging & inefficient to describe computations of Fully Homomorphic Encryption (
#FHE
) using Fields structures.
FHE is based on "Ring" algebraic structures, not Field, making working with both difficult, slow & overall not a fun time.
Calling all
#blockchain
community 🗣️
Are you ready to play
#FHE
FOMO ?
Let's dive deep into an exciting on-chain game on ETH-Sepolia testnet.
Here is a tutorial about how to play 👇
1/8
5/7 We took a new approach and built SNARG for Ring structures - This lowers computational requirements drastically and means faster proof generation, achieving publicly verifiable FHE for AI inference.
How does this look in action? 👇
We've joined
@shared_security
alongside other talented builders, researchers and integrators in the restaking ecosystem 🤝
We look forward to discussing shared security and restaking on July 8th in Belgium! 🇧🇪
@theSightAI
is accelerating restaking as a Founding Member of the Shared Security Alliance! 🌐
Join us at the Symposium on July 8th to hear their perspective on building with shared security! ➡️
Traditional RNG methods have some limitations such as:
1. Predictable: It is possible to predict the random number before the actual RNG transaction has been confirmed;
2. Manipulation: The
#Oracle
node may set a target value and claim that it is generated at random.
2/4 👇
3/6 vFHE enables secure & private computations on encrypted data, ensuring that results can be independently verified without exposing sensitive data.
The guarantee that web3 AI model outputs are correct & have not been tampered with opens the floodgates for AI dApp development.
5/6 Our lightweight prove-verify protocol also lowers computational requirements considerably, with the compute node running FHE model inference, generating SNARG proof.
The public verifier then uses an on-chain contract to prove validity.
Then it's time to explore it here:
This tech keeps crucial game elements confidential and secure showcasing real-world
#FHE
applications.
Happy
#coding
!💻
2/7 When it comes to ZK + FHE, today's zkSNARK schemes use "Field" algebraic structures for efficient generation of zero knowledge characteristics.
But those currently utilising zkSNARK for FHE verifiability are hitting a pretty big snag.
How so? 👇
Step 5: Place Your Bet 🎲
• Select 0.001, 0.005, or 0.01 $ETH
• Sign the transaction
• View pool details:
Winner (if any)
Encrypted target
Total pool sum
Your deposit
Remember: One bet per pool!🎲
7/8
2/ Greater Security for AI Models 🔒
vFHE means AI dApps can ensure data remains encrypted throughout AI processing lifecycles.
This not only secures the data but also the AI models themselves, as the models never directly interact with unencrypted data, reducing risk vectors.
FHE ensures privacy and security in blockchain systems, allowing computations on encrypted data without decryption.
Explore the history of
#FHE
, it's applications, and how it can enable millions of new businesses to employ blockchain systems:
Step 6: Check Results 🏆
• After the game ends, you can view completed pools
• See the winner's address and decrypted target
Good luck and have fun playing
#FHE
FOMO! May the odds be in your favor! 🍀
For more info, visit:
1/ Enhanced Trust & Transparency 🤝
VFHE adds a layer of trust to dApps by allowing parties to verify the correctness of computations done on encrypted data.
This is crucial in decentralized environments where trust needs to be established without relying on a central authority.
2/6 Data verifiability is paramount for the future of web3 AI, where multiple parties may have different interests.
Verification of output ensures web3 AI computations are performed accurately & honestly, with no need for trust in a central authority.
Enter verified FHE (vFHE).
Sight AI Co-Founder Jiapeng joins a stacked panel at the FHE Summit today, hosted by
@FHEOnchain
!
If you're in Brussels for
#ETHCC
this is one not to miss.
Salam Alaikum Dubai!
Our founding team is excited to explore the future of AI, Depin, and Web3 at this event. We are grateful for the invitation and look forward to connecting with fellow DeAI enthusiasts.
Come meet our team if you're here!🤩🌟