In its latest attempt to address AI security and privacy concerns, Microsoft provided new details about the underlying architecture for its upcoming Recall feature, which promises to let Windows users quickly locate information they’ve previously viewed on their PCs with help from artificial intelligence.
Snapshots taken in Recall will be encrypted and accessible only if the user has been authenticated using the Windows Hello biometric sign-in feature, the company said. In addition to making Recall opt-in, the company now says it will give users the ability to completely remove Recall from their machines if they choose.
Those were among a series of details in a Microsoft post Friday morning.
Recall, which takes advantage of advanced Neural Processing Units (NPUs) in new Windows PCs, was announced in May as the flagship feature in Copilot+ PCs from Microsoft and other computer makers. The first Copilot+ PCs were released in June without Recall, pending future software updates.
Recall has drawn close scrutiny from security researchers and legislators, especially in light of Microsoft’s pledge to prioritize security over new features following a series of high-profile breaches exploiting flaws in its software and cloud services. Microsoft gave a broader update about its security initiatives earlier this week.
“Recall’s secure design and implementation provides a robust set of controls against known threats,” wrote David Weston, Microsoft’s VP of Enterprise and OS Security, on Friday. “Microsoft is committed to making the power of AI available to everyone while retaining security and privacy against even the most sophisticated attacks.”
Microsoft said previously that it plans to release the Recall feature initially through its Windows Insider program in October.
