How to Avoid AI Labelling for Your Edited Images on Meta Platforms?
Image Source: https://www.instagram.com/jackylee.ai/
Meta has announced a significant update to how AI-generated and AI-edited content is labeled across its platforms. The shift from the 'Made with AI' to 'AI Info' label aims to provide users with more clarity about the content they encounter. This change was spurred by feedback, including notable instances where users like former White House photographer Pete Souza experienced mislabeling of content. Souza's decades-old film photo was mistakenly tagged as AI-generated, highlighting the need for this update.
The Trigger Behind Mislabeling
The issue of mislabeling seems to stem from the metadata that editing tools like Adobe Photoshop attach to images. Meta's expanded AI content policies led to situations where even traditional photos edited with such tools were tagged as AI-generated. This has sparked a debate between Adobe and Meta regarding the interpretation and presentation of metadata, with Adobe advocating for labels only when content is entirely AI-generated.
Industry Responses and Recommendations
Adobe, through Andy Parsons, the senior director of the Content Authenticity Initiative, has expressed concerns about the broad application of AI labels. Parsons argues that AI enhancements used to achieve aesthetic improvements should not necessarily trigger an AI label unless the content is fully generated by AI. This stance points to a broader industry challenge of balancing technological integration with accurate content representation.
Rollout and User Interaction
Meta plans to implement the new 'AI Info' label across all its platforms, starting with mobile apps followed by web interfaces. Users interacting with the label will find the same detailed explanation as before, clarifying why the label was applied. This includes whether the image was fully generated by AI or merely edited using AI technologies.
The Future of Metadata and AI Labeling
Despite advancements in metadata tagging technologies like C2PA, distinguishing between AI-generated and real images remains complicated. The ideal future where this differentiation is clear and straightforward is still out of reach, reflecting ongoing challenges in digital content authenticity.
Avoiding the 'AI Info' Label
For users concerned about their content being labeled as AI-influenced, there are workaround strategies. Removing EXIF data from image files or creating a new Photoshop file to strip previous metadata are effective methods to prevent Meta's AI detection tools from mislabeling content. These steps, while additional, offer users more control over how their content is perceived and labeled.
Spotlight on Jacky Lee
Jacky Lee, a renowned landscape and architectural photographer, has embraced AI in creating virtual artworks of "camera buildings" in a cityscape. His professional experience in building surveying and project management lends a realistic quality to these virtual constructions. Despite not having the 'AI Info' label on his uploads, Lee chooses to mark his AI-generated images explicitly, demonstrating responsible AI artistry.
The Ethical Dimension of AI Art
Jacky Lee's approach highlights an essential aspect of AI in art: transparency. By voluntarily marking his AI-generated artworks, Lee sets a standard for ethical AI use in the creative industries. This practice not only informs the audience but also fosters trust and sets a precedent for other AI artists to follow, ensuring that viewers are aware of the nature of the content they are engaging with.
Source: https://www.theverge.com/2024/7/1/24190026/meta-instagram-facebook-made-with-ai-info-label-metadata
