Facebook — Policy Change

majorAnti-PrivacyMay 22, 2023 → Sep 27, 2023Policy Change

Executive Summary

Meta launched consumer-facing generative AI assistants (Meta AI) and established that public posts, photos, and content — but not private messages — could be used to train Meta's AI models. Users were given opt-out mechanisms, though the process was not straightforward.

Post LinkedIn Reddit Email

data-collection ai-training content-licensing

What Happened

On September 27, 2023, Meta announced the launch of Meta AI, a generative AI assistant available across WhatsApp, Messenger, and Instagram, with plans to expand to Ray-Ban Meta smart glasses and Quest 3. The company established that public posts, photos, and content could be used to train Meta's AI models, though private messages would not be included. Meta AI is powered by technology from Llama 2 and has access to real-time information through a search partnership with Bing, and can generate images using Meta's Emu model.

Who Is Affected

All users of Meta's platforms including Facebook, Instagram, WhatsApp, and Messenger are affected, with their public posts and content potentially used for AI training. The initial rollout of AI features like AI stickers is limited to select English-language users over the following month. Users were provided opt-out mechanisms, though the process was described as not straightforward.

Why It Matters

This policy change represents a major shift in how one of the world's largest social media companies uses user-generated content, affecting billions of users who share public posts across Meta's platforms. The establishment of public content as training data for AI models sets a significant precedent for how tech companies may use existing user content without explicit prior consent. The lack of a straightforward opt-out process raises concerns about meaningful user control over their data.

What You Should Do

Review your privacy settings across all Meta platforms and change any public posts to private or friends-only if you do not want them used for AI training. Locate and use the opt-out mechanisms provided by Meta, despite them being described as not straightforward. Consider limiting what you share publicly on Meta platforms going forward, and review Meta's AI data usage policies to understand what content is being used and how.

AI-Assisted

Event summaries are generated by Claude AI from verified sources and reviewed by humans before publication.

Sources

Related Events