Skip to main content

Share story

AI Products

Google outlines encrypted cloud memory for its AI assistants

Google says it plans to add persistent cloud memory to its Private AI Compute platform so an assistant can retain context across a person's devices. Under the proposed design, encrypted information would stay in dedicated cloud storage while the keys needed to unlock it would remain on the person's devices. For a request, an authenticated encrypted connection would send data into an isolated cloud environment, where it would be decrypted temporarily, updated and encrypted again. Google describes how the system is intended to work; the announcement does not establish that the memory service is available to users.
Sources
In this story
Published by Tech & Business, a media brand covering technology and business. This story was sourced from Google DeepMind News and reviewed by the T&B editorial agent team.
Back to Newswire
Keep reading
Full wire
AI Products
AI Products

Google separates school Search and Gemini controls after backlash

Google has backed away from a setting that required schools to disable Google Search when turning off Gemini, EFF reports. Google clarified that Gemini, Gemini notebook and Search will have separate controls and will be off by default.

AI Policy
AI Policy

Meta indefinitely delays commercial terms expanding AI training access

Meta has indefinitely delayed commercial terms that would require business account users to permit AI training on external websites linked on Facebook, Ars Technica reports. A Meta spokesperson said existing terms remain in effect while the company engages with stakeholders.

AI Products
AI Products

Ramp finds AI spending growth slowed despite rising usage

AI token spending grew 1.5% across August and September despite rising usage volumes, according to a new Ramp report described by TechCrunch. July spending had grown 42% from the previous month.

AI Products
AI Products

Microsoft winds down rented Oracle AI computing capacity

Microsoft is switching off computing capacity rented from Oracle after declining to renew a 2023 deal because the hardware used older Nvidia chips, Bloomberg reported, citing sources.