Project Update August: What’s Next for iExec AI?
Welcome to our latest community update! As we continue to push the boundaries of innovation at iExec, we’re excited to share our progress on our mission to build, own, and monetize in Web3. Our team is working tirelessly to deliver cutting-edge solutions that meet the evolving needs of the Web3 space.
In this update, we’ll dive into the advancements we’re making in the blockchain/AI space, we’ll explore Privacy Pass, and highlight events we’ll be attending/hosting to help shape the future of decentralization. Whether you’re a developer, an end-user, or simply an enthusiast in the community, there’s something for everyone. Let’s take a closer look at how iExec is setting the stage for the next generation of secure, scalable, and privacy-preserving technologies.
AI and iExec
AI is one of the hottest tech trends in the last decade, and with good reason. As you know, iExec is exploring the integration of AI with blockchain technology, tackling some of the most pressing challenges in data confidentiality, trust, scalability and monetization. Our work isn’t just theoretical; it’s about providing the infrastructure dev tools that make it easy to create real-world applications that empower AI model owners to monetize their assets while keeping data secure and private.
Our focus is on enhancing AI applications by leveraging the strengths of blockchain and confidential computing, using off-chain computing solutions that bring privacy and monetization of AI. There are numerous challenges to address, like data privacy, scalability, and the notorious “Black Box” problem in AI. The ultimate goal is to enable AI model owners to monetize their assets while keeping full control and confidentiality intact.
Through strategic collaborations with industry giants like Intel, Microsoft, and Google, we’re making strides in integrating AI with blockchain in ways that prioritize security, transparency, and trust. Partnering with industry heavyweights like Intel and playing a key role in initiatives like the Intel AI Builder Program, iExec has earned its stripes as a leader in trusted computing with blockchain.
As AI continues to evolve, iExec’s solutions are uniquely positioned to empower businesses that see the potential of AI but are held back by confidentiality or privacy concerns. iExec provides infrastructure that enables the deployment of AI while safeguarding sensitive data. This makes iExec an ideal solution for industries such as healthcare, fintech, and cybersecurity, where data privacy is paramount.
MACHINE LEARNING
We’re making exciting strides in how users can contribute to machine learning while keeping their data confidential. Our AI research team is working on a framework that allows users to encrypt their data and securely contribute it to the machine-learning process. To demonstrate the potential of this approach, we’re looking into developing use-case demos, such as image description and image generation, which showcase how inference data, models, and results can be kept private.
Stay tuned as we continue to push the boundaries of secure AI development.
Our research includes experimenting with GPU worker pools, and the results so far are very promising. Our framework shows strong potential for decentralized training, which could lead to better-trained models on higher-quality data. This could address a significant issue: institutions or individuals often hesitate to share valuable, confidential data for model training. With our framework, they can contribute securely and be rewarded without exposing their data. We’re also in contact with a university that’s interested in applying this framework in healthcare, where sharing patient data is particularly sensitive.
Looking ahead, we’ve got research on a neurology use case in the pipeline and are considering expanding our team to bring in additional engineers. These new team members would work on AI-centric proof-of-concepts. We met with Intel earlier this month where we discussed TDX + AI, confirming they are actively working on supporting TDX + H100
AI AND FEDERATED LEARNING
At the Confidential Computing Summit earlier this year, Anthony and Joey discussed their pioneering work in the intersection of AI, Federated Learning (FL), and Confidential Computing (CC). As we generate vast amounts of data, it becomes a huge challenge to train powerful AI models like Large Language Models (LLMs) while maintaining data privacy. Federated Learning offers a solution by allowing multiple devices to collaborate on training a model without sharing the data directly. However, this process still faces security risks, as model updates exchanged between devices can potentially expose sensitive information.
To address these challenges, iExec introduced the concept of Delegated Federated Learning. In this framework, users contribute encrypted data while remote worker nodes, equipped with Trusted Execution Environments (TEEs), handle the Federated Learning process. This ensures that data remains confidential throughout: workers never access the raw data or the shared model. The iExec team showcased this framework's initial design and proof-of-concept, leveraging blockchain and advanced TEE technologies like Intel SGX, TDX, and Nvidia Hopper Confidential Compute. This approach has the potential to revolutionize Confidential AI, enabling secure and privacy-preserving AI development on a decentralized platform.
Joey is now working on integrating this into the iExec stack, much like any other Dapp. This process has already highlighted some limitations in our stack, and his findings will guide our development efforts to make the iExec platform AI-optimised.
NEW PARTNERSHIPS
Lobster is a DeFi platform committed to secure and user-friendly crypto experiences. They’ve integrated iExec to enhance the privacy of its communications! Users can receive updates without revealing their email addresses by using iExec’s Web3Mail tool into the Lobster newsletter. This integration leverages iExec’s advanced encryption and confidential computing technologies, ensuring that users’ personal data remains secure while they stay informed. On the flip side, this collaboration underscores Lobster’s dedication to pushing the boundaries of decentralized finance while maintaining top-tier privacy and security for its community.
iExec PRIVACY PASS
Privacy continues to gain importance in both current international events and the rapidly evolving world of decentralized finance. Against this backdrop, iExec has launched its latest initiative, Privacy Pass. This campaign not only emphasizes the significance of privacy in the Ethereum ecosystem but also rewards users with RLC for participating in privacy-focused campaigns. As the initiative progresses, iExec is committed to refining and expanding its offerings to better serve the community.
This campaign allows users to earn RLC tokens simply by reading privacy-centric emails. The campaign is still in its beta phase, with ongoing improvements based on user feedback. Privacy Pass aims to attract more end-users and partners to create a strong network effect. Each campaign is managed through a dashboard, where users can track their participation and subscribe to new campaigns as they launch. Details on future partnerships remain under wraps, but users can expect more marketing efforts and campaigns to follow. Stay tuned to iExec’s official channels for updates.
iExec VOUCHERS
Attention builders! iExec is preparing to soft-launch a game-changing tech solution designed to simplify your interaction with Web3 protocols. This innovation will help developers know their development costs upfront, eliminating surprises and enhancing your building experience. We’re starting with a closed beta, offering a select group of builders free access to premium resources and support. Stay tuned for more details coming soon!
UPCOMING EVENTS
Over the next few months, iExec is excited to participate in a series of events and workshops aimed at empowering developers and advancing the Web3 ecosystem. Here’s what’s coming up:
- Encode Boot Camps and Workshops: We’re hosting workshops on iExec developer tools for participants in Encode Club’s Zero-Knowledge (ZK) and Scaling Bootcamp cohorts. Additionally, we’ll support Encode Club’s series of six bootcamps and an online hackathon running from September to December, providing valuable opportunities for developers to expand their skills in the Web3 space.
- ETHLisbon: We’re excited to sponsor ETH Lisbon Hackathon for a third time, empowering devs to ger creative with iExec’s easy-to-use tools for data protection and monetization in #DeCC!
- ETH Naples: Workshops where we’ll delve into techniques for monetizing and protecting data using developer tools.
- Token2049: iExec will be attending Token2049 and co-organizing events focused on promoting the Decentralized Confidential Computing (DeCC) narrative, fostering discussions and collaborations within the blockchain community.
- ETH Rome Hackathon: We’re proud to sponsor the upcoming ETH Rome hackathon, encouraging innovation and supporting developers as they build the next generation of decentralized applications.
- DevCon Bangkok: Meet the iExec team at DevCon Bangkok, where we’ll engage with the global Ethereum developer community, share insights, and explore new possibilities in decentralized computing.
Stay tuned for more updates and detailed information about each event. We look forward to connecting with you — let us know which event you’re most excited to attend!
As we move forward, we’re excited to share these developments with you and invite you to stay engaged as we explore new possibilities and push the boundaries of what’s possible in the Web3 space. Thank you for being an essential part of the iExec community.
Source : iExec - Sep 1, 2024