Microsoft's 'Recall' Feature Will No Longer Be a Default Setting Due to Security Concerns

Microsoft's controversial new AI Recall feature will no longer be activated by default for Windows users, the company announced.

Recall, which was unveiled in May and is currently in preview, records every action taken by a user to help create more personalized queries in its AI-powered Copilot. While the new feature has been designed to improve the Copilot experience, some in the security community began to express possible concerns.

In an article posted to Medium (soft paywalled), security expert Kevin Beaumont broke down how, with just a few lines of code, Microsoft's newest Copilot feature could allow attackers to steal a user's entire PC usage history.  In his analysis, he said that, while the new feature would be useful to many, Microsoft's security track record would make this feature too risky for public use.

"I think it's an interesting entirely, really optional feature with a niche initial user base that would require incredibly careful communication, cybersecurity, engineering and implementation. Copilot+ Recall doesn’t have these," Wrote Beaumont. "The work hasn’t been done properly to package it together, clearly."

Responding to feedback from Beaumont and others, along with disabling the feature by default, Microsoft said it is adding an additional "enrollment" level to use the service and enhancing Recall's security with "just in time" decryption, which uses Windows Hello Sign-In Security (ESS).  

It also laid out the current security settings for the Copilot feature. Recall, locally stores and processes snapshots without relying on Internet or cloud connections. These snapshots are never shared with Microsoft or other entities, and per-user encryption ensures privacy even on shared devices.

According to Microsoft, users will be clearly notified when Recall is saving snapshots, with icons visible on the taskbar and system tray. Digital rights managed content and InPrivate browsing data are excluded from snapshots.

Finally, users can manage their snapshots with options to pause, filter or delete them at any time, maintaining control over what is saved. For enterprise users, IT administrators can disable the snapshot-saving feature, but cannot enable it on behalf of users.

Microsoft also said the line of Copilot+ PCs, also announced last month, will come with additional security features, including hardware-level baked-in security, Microsoft Pluton security processor and Windows Hello ESS turned on by default.

"We are on a journey to build products and experiences that live up to our company mission to empower people and organizations to achieve more, and are driven by the critical importance of maintaining our customers’ privacy, security and trust. As we always do, we will continue to listen to and learn from our customers, including consumers, developers and enterprises, to evolve our experiences in ways that are meaningful to them," wrote Pavan Davuluri, corporate vice president, Windows + Devices. 

About the Author

Chris Paoli (@ChrisPaoli5) is the associate editor for Converge360.

Featured

  • digital graph

    Building Genuinely Data-Informed School Districts

    Schools today generate enormous amounts of data. The difference between collecting it and using it effectively often comes down to quality, access, and context.

  • large cloud icon on the right in an abstract world above a polygon with a dark blue background

    Cloud Security Alliance Expands Agentic AI Governance Work

    The Cloud Security Alliance (CSA) has announced a series of CSAI Foundation milestones aimed at securing what it calls the agentic control plane, including a new catastrophic risk initiative, CVE Numbering Authority authorization, and the acquisition of two agentic AI specifications.

  • circuit patterns

    Anthropic Intros Lower-Cost Claude Sonnet 5

    Anthropic has launched Claude Sonnet 5, positioning the model as its most autonomous mid-tier offering to date and a lower-cost alternative to its flagship Opus 4.8 system. The company said the model can plan multi-step tasks, operate tools such as browsers and terminals, and complete agentic work at a level that previously required larger and more expensive models.

  • digital lock with circuit patterns

    IBM Adds New AI-Powered Cybersecurity Tools

    IBM has unveiled an expanded portfolio of AI-powered cybersecurity products, positioning the company to compete more aggressively in a rapidly evolving market where enterprises are increasingly turning to artificial intelligence to defend against automated cyber threats.