Meta AI Smart Glasses Footage Reviewed by Kenyan Contractors
In February 2026, investigations revealed that Meta’s AI-powered smart glasses, marketed as Ray-Ban Meta, captured video and images—including intimate, medical, and sexual moments—without explicit user consent. This footage was then routed to low-paid human contractors in Kenya for annotation purposes, raising significant privacy concerns. The contractors, employed by Sama, a subcontractor based in Nairobi, reported viewing sensitive content such as nudity, bathroom visits, sexual activity, and bank card details. Some individuals recorded appeared unaware they were being filmed, while others had their faces left visible despite Meta’s claims of anonymization filtering.
The disclosure of this practice revived earlier controversies, including the 2019 incident where voice-assistant recordings were sent to human reviewers for grading. Meta’s smart glasses were positioned as a privacy-focused device, with the company emphasizing in 2023 that ‘Ray-Ban Meta smart glasses are built with privacy at their core.’ However, the investigation by Swedish newspapers Svenska Dagbladet and Göteborgs-Posten contradicted these assurances, exposing a systemic failure in data handling and consent mechanisms.
Some recordings were triggered by AI features designed to assist users, such as visual queries or context-aware interactions. For example, moments involving changing clothes, medical consultations, or private conversations were inadvertently or intentionally included in the data sent for annotation. Workers described seeing individuals in vulnerable states, such as using the bathroom or engaging in intimate activities, often without their knowledge or consent.
The revelation prompted immediate backlash, including lawsuits and regulatory inquiries. A class-action lawsuit was filed against Meta, alleging violations of privacy laws and misrepresentation of the glasses’ capabilities. The Office of the Data Protection Commissioner (ODPC) in Kenya launched an investigation to determine whether the use of this footage complied with local and international data protection regulations. Civil society groups, including the Kenyan digital rights organization Oversight Labs, also called for an independent probe into the matter, questioning whether the data was being used to train artificial intelligence models without proper safeguards.
Meta responded to the controversy by terminating its contract with Sama, which resulted in the abrupt dismissal of 1,108 workers in Kenya. The company stated that reviewers were only exposed to data that users had explicitly shared with AI features, though this claim was contradicted by worker testimonies. Meta also tightened its disclosure policies, though the damage to trust and the legal repercussions remained unresolved.
Privacy Violations and Systemic Failures
The core harm in this incident was the compromise of privacy, categorized under the Domain Taxonomy of AI Risks as ‘Compromise of privacy by obtaining, leaking, or correctly inferring sensitive information.’ The AI system in question—Meta’s smart glasses—was designed to capture and process visual data in real time, but the lack of robust safeguards allowed for the exposure of highly personal moments. Workers reported instances where faces were not adequately blurred, further exacerbating the privacy violations. Some individuals recorded appeared to have no awareness that their activities were being documented, undermining the principle of informed consent.
The company’s reliance on third-party contractors for sensitive tasks introduced risks that were not adequately mitigated. Sama, the Kenyan firm responsible for annotation, employed workers on precarious contracts with minimal oversight, creating an environment where ethical concerns could be overlooked.
Regulatory and legal responses to the incident underscored the inadequacies in existing frameworks for governing AI-driven data collection. The ODPC’s investigation aimed to clarify whether Meta’s practices violated Kenya’s data protection laws, while the class-action lawsuit sought damages for affected users. However, the global nature of Meta’s operations and the fragmented regulatory landscape posed challenges to comprehensive accountability. The incident served as a case study in how AI systems, when deployed without sufficient transparency or ethical oversight, can lead to unintended and harmful consequences.
Impact on Workers and the AI Economy
The termination of 1,108 Sama workers following Meta’s contract cancellation drew attention to the precarious nature of jobs in the global AI economy. These workers, primarily based in Nairobi, were part of a larger trend where tech companies outsource labor-intensive tasks—such as data annotation—to regions with lower wages and weaker labor protections. The sudden job losses highlighted the fragility of these positions, which often lack job security, fair compensation, or avenues for recourse when ethical violations occur.
Workers who spoke to investigators described the psychological toll of reviewing graphic content, including footage of individuals in distressing or intimate situations. While Meta’s decision to end the contract may have been a response to public pressure, it did little to address the underlying issues of exploitation within the AI supply chain. The incident raised questions about corporate responsibility in ensuring fair and safe working conditions for contractors involved in AI development.
The controversy also sparked discussions about the broader implications for AI ethics and governance. Critics argued that Meta’s smart glasses, like other AI-powered devices, require stringent privacy protections and independent audits to prevent similar breaches. The incident reinforced the need for clearer regulations on data annotation practices, particularly when sensitive or personal data is involved. Without stronger safeguards, the risks of privacy violations and worker exploitation in AI-driven industries are likely to persist.
- Privacy violations: Footage included sexual, medical, and bathroom moments, with some individuals unaware they were recorded or faces left visible despite filtering.
- Systemic failures: Meta’s reliance on third-party contractors (Sama) for annotation introduced ethical and operational risks, including lack of oversight and worker exploitation.
- Regulatory and legal consequences: Lawsuits and ODPC investigations were launched, though fragmented global regulations limited comprehensive accountability.
- Worker impact: 1,108 Kenyan contractors were dismissed after Meta terminated the Sama contract, exposing vulnerabilities in the AI labor market.
- AI ethics concerns: The incident highlighted gaps in data protection, annotation practices, and corporate responsibility in AI development.
Sources
Incident summary from the AI Incident Database, used under CC BY-SA 4.0.