Meta executives give conflicting accounts of NameTag face recognition system

In June, Meta ↗ 1.44% dismissed reports about its facial recognition system as sensational and declared the feature "does not exist." Five weeks later, its chief technology officer openly described the same system by name, explained how it works, and called it useful. The contradictions have drawn renewed scrutiny from privacy advocates and journalists seeking clarity on whether the company is building biometric surveillance tools for its smart glasses.

From Denial to Description

The confusion traces back to a June 4 WIRED investigation that uncovered dormant facial recognition code — internally called NameTag — embedded in Meta's AI companion app, which has been installed on more than 50 million phones. The code had been integrated into the app through updates beginning in January 2026 without any public disclosure.wired+2

After the report, Andy Stone, Meta's vice president of communications, rejected WIRED's findings, asserting the company could not answer questions about how the system would function because "the feature does not exist". Andrew Bosworth, Meta's CTO, went further, calling the reporting "incredibly misleading" and "absolutely dishonest". Meta then stripped the NameTag code from the app within a day of the report.facebook+3

Yet in a July interview with journalist Nicholas Thompson, Bosworth discussed NameTag at length as a real feature with a clear design and purpose. "That's what we call the NameTag feature," Bosworth said, describing it as a locally encrypted system that would remember the faces of people a user meets while wearing Meta's smart glasses. He pitched it as solving "the cocktail party problem" of forgetting names, and suggested accessibility benefits for blind and low-vision users.gizmodo

Questions of Transparency

The gap between Stone's categorical denial in June and Bosworth's detailed description in July has raised questions about Meta's internal coordination and public accountability on sensitive AI features. WIRED published a follow-up on July 14 noting the contradictions, asking whether a software feature "exists" once its code has been deployed to millions of devices.wired+1

Privacy advocates have pointed to Meta's history with biometric data — the company previously built a database of 1 billion faces through Facebook before deleting it after lawsuits from Texas and Illinois. Meta spokesperson Ryan Daniels maintained in June that "we are not building a central face database," describing all work as purely "exploratory".engadget+2

The episode underscores a recurring tension: Meta continues to develop powerful biometric tools while simultaneously minimizing their readiness, leaving the public to reconcile official denials with the code already sitting on their phones.