Google Plans Encrypted Cross-Device Memory for Private AI Compute
Summary
Google plans to add persistent memory to Private AI Compute, using encrypted storage, device-held keys and secure cloud enclaves. The company has published technical details and invited external review; example uses are not yet available features.
Key points
- The planned memory layer would let AI assistants retain context across sessions and devices; the platform has been stateless until now.
- Google says information will be stored in encrypted storage, with the keys held exclusively on users’ devices.
- When context is needed, an authenticated, end-to-end encrypted connection will deliver it to a secure cloud enclave for temporary processing.
- Google says the enclave will save new context and encrypt it again after processing.
- Google has published an updated technical brief, a tamper-proof public record of server software and results from an independent audit.
- Devices will be able to verify the software is authentic and unaltered before sending personal data, and Google has invited privacy-community review.
Article Details
- Event Type
- Google announced plans to add persistent, server-side memory to Private AI Compute.
- Impact
- The planned feature would let AI assistants retain context across sessions and devices. Google says data will be encrypted, with keys held on users’ devices, and processed in secure cloud enclaves so Google and others cannot access it. The described cross-device scenarios are potential future uses, not currently available features.