Microsoft Recall Faces Renewed Scrutiny Over AI Privacy Risks

Microsoft's Recall faces new privacy criticisms after researchers bypass updated security protections.

Microsoft's Recall feature—intended to boost productivity through AI-powered search of a user's past desktop activities—remains a flashpoint for privacy and cybersecurity concerns, even after a recent overhaul. Recall captures screenshots of a user's desktop every few seconds, organizing them for retrieval using natural language queries, and operates exclusively on Copilot+ PCs meeting high hardware requirements.

• The revamped Recall, relaunched in April 2025, touts robust safeguards like AES-256-GCM encryption, Windows Hello biometric authentication, and a Virtualization-Based Security (VBS) enclave.
• Despite these advancements, cybersecurity researcher Alexander Hagenah demonstrated with his TotalRecall Reloaded tool that injecting code into the AIXHost.exe process grants access to decrypted screenshots and metadata, bypassing the promised protections. Hagenah likened it to: "The vault door is titanium. The wall next to it is drywall."
• Microsoft responded that this exploit operates "within the current, documented security design of Recall," and opted not to take further action at this time.
• The episode adds to a growing chorus of criticism over rapid AI integration, with Mozilla's VP of Global Policy, Linda Griffin, noting a pattern of "going too far without user consent" from Microsoft, after some Copilot features were rolled back following backlash.

For legal operations and corporate counsel, the persistent vulnerabilities highlight the need for careful risk analysis—and may signal future challenges in ensuring data protection when adopting embedded AI technologies.

Yes, but: Microsoft has not indicated any immediate plans to address the newly exposed Recall vulnerability, leaving organizations to weigh risks of adoption.