Meta Introduces Labels for AI-Generated Content on Facebook and Instagram

Meta is set to implement new content policies that will label AI-generated material across its platforms, including Instagram and Facebook, beginning in May. This initiative aims to enhance transparency by clearly marking user-generated content produced using artificial intelligence with a “Made with AI” label, offering users clearer context about the nature of the media they encounter.

This labeling strategy stems from Meta’s commitment to address the rising concerns regarding manipulated content. The company plans to leverage “industry-shared signals” to identify AI-altered images, but it will also allow users to voluntarily disclose if their content has been AI-generated. To further protect users from potential misinformation, any content that is deemed likely to mislead will receive a more prominent labeling designation.

Earlier, Meta took down manipulated content to maintain trust, but starting in July, the platform will cease removing such content for a two-month transition period. According to Monika Bickert, Meta’s vice president of content policy, this timeframe will help users adapt to the new self-disclosure system. Despite this relaxation, manipulated media that contravenes the platform’s Community Standards—specifically related to issues like voter interference, bullying, harassment, and incitement of violence—will still be removed.

Meta’s policy update responds to recommendations from its Oversight Board, which had deemed the previous guidelines surrounding AI content too limited, as they focused solely on video. The board argued that the original framework was not comprehensive enough to address the broader implications of AI-generated content. Their concerns were highlighted during a review of altered footage involving public figures, including President Joe Biden, where the board emphasized the importance of maintaining freedom of expression while ensuring transparency.

In response to the Oversight Board's feedback, Meta has expanded its moderation framework to encompass all forms of AI-generated content, including video, audio, and images. Bickert affirmed the company's agreement with the board’s perspective, stating, “We agree with the Oversight Board’s recommendation that providing transparency and additional context is now the better way to address manipulated media.” By retaining this content and adding labels, Meta aims to foster a more informed user community without unnecessarily limiting users' freedom of speech.

To gauge user sentiment, Meta recently surveyed 23,000 individuals and found that a significant 82% favored the implementation of labels for AI-generated content. This insight reflects a growing demand for clarity in digital media. Additionally, the platform's image generation feature, currently available only to U.S. users, includes an “imagined with AI” watermark on the images it produces, further reinforcing its commitment to transparency in the realm of artificial intelligence.