Meta to Label AI-generated images on Facebook, Instagram, and Threads


Meta today announced its collaboration with industry partners to establish common technical standards for recognizing AI-generated content, encompassing both video and audio formats.

New Labeling System for AI-Generated Content

To address concerns around transparency, Meta is working on a labeling system to denote AI-generated photorealistic content, including images created with tools beyond Meta’s own.

This initiative involves applying “Imagined with AI” labels to Meta AI-generated images, with plans to extend this to content produced by other companies’ tools.

Establishing Standards and Tools

Collaborating with industry counterparts through forums like the Partnership on AI (PAI), Meta aims to develop common standards for identifying AI-generated content.

Their approach aligns with PAI’s best practices, utilizing IPTC metadata and invisible watermarks. Meta is also developing tools capable of identifying invisible markers at scale, facilitating labeling across various platforms.

Challenges and Future Prospects

Despite progress, identifying AI-generated audio and video content remains a challenge. Meta plans to introduce features allowing users to disclose AI-generated video or audio content, ensuring appropriate labeling.

Additionally, Meta is exploring advanced watermarking technologies like Stable Signature to enhance detection capabilities.

Looking Ahead and Community Standards

As the prevalence of AI-generated content grows, Meta emphasizes the importance of ongoing dialogue with industry peers, governments, and civil society. While setting out initial steps for content labeling, Meta remains receptive to feedback and committed to continuous improvement.

Furthermore, Meta underscores its commitment to upholding Community Standards, leveraging AI to combat harmful content effectively.

Availability

In the coming months, Meta plans to roll out labeling for AI-generated images on Facebook, Instagram, and Threads, marking a step forward in enhancing transparency for users.

Announcing the update, Meta posted:

As a company that has pioneered AI development for over a decade, it’s immensely inspiring to witness the surge of creativity unleashed by our latest generative AI tools, such as our Meta AI image generator, enabling users to craft images from basic text prompts.

With the line between human and synthetic content becoming increasingly indistinct, there’s a growing demand for clarity regarding its origin. Many individuals encounter AI-generated content for the first time, and our users have emphasized the importance of transparency in navigating this new landscape. Hence, it’s crucial for us to ensure users are aware when the photorealistic content they encounter has been generated using AI.