What happened
On September 23, 2026, Google's Private AI Compute Team outlined plans to add private, server-side memory to its Private AI Compute platform, which already let users run demanding jobs inside cloud enclaves isolated at the hardware level. The planned memory is a lasting feature, described as working much like a protected storage locker hosted in the cloud. The announcement did not say who will be eligible to use it.
Why it matters
The planned memory layer would store data in the cloud while the cryptographic keys required to open it stay solely on users' own devices. Google also stated that it will release an unalterable public log of its server software, along with findings from an outside audit conducted by a major cybersecurity company.
Read the original at Google DeepMind