The updated label is a response to AI being used to edit and manipulate images, in addition to generating them.

The new ‘AI info’ label that replaces the ‘Made with AI’ label. (Image Credit: Meta).

New Delhi: Meta has evolved its ‘Made with AI’ label, and replacing it with an ‘AI info’ label, that provides additional contextual information on how AI tools were used to produce the image. Meta had earlier introduced the ‘Made with AI’ label after the proliferation of content on Instagram, Threads and Facebook that were created using generative AI tools. Now, Meta has seen the requirement for providing additional details, considering there are AI tools that allow for image editing and manipulation as well, as against just generating new images based on text prompts.

The changes are being rolled out based on the recommendations of Meta’s Oversight Board, along with a policy review process that includes public opinion surveys and expert consultations. Meta is working closely with the industry to develop the signals it uses to recognise content created using AI. Meta is expected to tag a wider range of audio, video and image content with the new label. Meta also recognised that their existing labels were not aligned with the expectations of people.

There were for example, retouching tools that made minor modifications to images using AI. These images were tagged with the ‘Made by AI’ label, which was just not true and did not paint the complete picture. Meta’s new approach is more transparent, and provides its users with additional information when necessary.

The ‘Made with AI’ label

Meta had introduced the ‘Made with AI’ label in April, after the proliferation of Generative AI content across its platforms. Before these Generative AI tools became common, synthetic images were rare on the platforms, with labelling necessary only for videos that were created or altered to make a person appear to say something they did not. At that time, fake videos were the overarching concern. There was a more emphasized label for content that Meta identified as being capable of potentially deceiving its audience.