To personalize AI, developers need user data while ensuring confidentiality. TEEs provide the means to process this information securely & efficiently.
@KalpitaMandal's guide on running Llama3.2 on Marlin's CVM showcases how TEEs enable this.
Sandboxed enclave images where dApps running within them don't affect the TEE's PCR values/measurements allowing easy verification of the TEE's security guarantees!
We're here at @EthereumDenver, ready to collaborate, innovate, and build. Super excited to connect with fellow hackers & builders and discuss all things AI and TEEs!
Marlin is working with @chromia to introduce a solution for AI agent memory persistence through TEEs. This marks the first implementation of secure, verifiable, and private memory management for AI models running onchain, enabling long-term context retention.
Marlin is working with @Aither_protocol to bring secure AI agent execution to Agent Virtual Machines (AVMs). This integration brings hardware-level security guarantees to AVM's containerized environment, ensuring safe and verifiable AI operations.
Marlin is working with @EniqAi to advance secure AI computation. ENIQ AI merges blockchain with AI-powered personalization, creating an educational ecosystem that requires secure, verifiable computation and data privacy guarantees.
In collaboration with @mellowprotocol and @symbioticfi we introduce the restaked POND vault to secure networks like @KalypsoProver - earn Mellow points, Symbiotic points and more!
Marlin Oyster now powers @Exponent_AI's Vector3 Network, delivering computational proof for AI scoring. The TEE platform runs language models in isolated environments, ensuring secure evaluation of LLM outputs.