Meta says it will allow its AI glasses users to opt out of having their “visual data” used to train its AI or shown to third-party contractors outside the US
Karissa Bell /Engadget:
Context & Ripple Effects
Meta’s glasses moved from a 2023 multimodal AI beta that analyzed photos toward a broader consumer AI device, while privacy safeguards became a recurring product issue: Meta removed code tied to an unreleased face-recognition system and added a camera shutdown measure for tampered privacy LEDs.
The new data choice sits alongside Meta’s planned Private Processing for assistant requests, which is designed to keep Meta from accessing certain user data. Together, the measures make data handling—not only camera hardware—a central part of the glasses proposition.
First-order effects
- AI-glasses users gain a stated choice to exclude their visual data from Meta’s AI training and from access by third-party contractors outside the US.
- Meta must separate opt-out users’ visual data from training and the specified contractor-access workflows, turning consent status into an operational control for its glasses service.
Second-order effects
- Meta’s privacy posture becomes more differentiated: Private Processing addresses Meta access for particular requests, while the opt-out governs training and specified third-party access, requiring users to understand separate data paths.
- Employers with sensitive-information concerns, including ICE, which barred staff from wearing the glasses, have a more granular consumer privacy control to weigh, though it does not itself change workplace-device policies.
Third-order effects
- If wearable AI makers pair visual assistants with granular consent and local/private processing, competition shifts toward proving where captured data travels and who may use it, not merely adding camera features.
- The policy points toward visual data becoming a separately governed training input, with access controls varying by purpose and contractor location rather than being treated as a single device-data category.
The trend: AI wearables are moving toward purpose-specific permission controls for the visual data that powers their assistants and model training.