Meta to label more content across photo, video & audio as ‘Made with AI’

The changes come after the Facebook content oversight board expressed concerns about the company’s existing approach to AI content, claiming for it to be insufficient. Currently, the company’s ‘transparency’ practice only covers images in which an individual is being manipulated. Such a policy was first introduced back in 2020, when “realistic AI-generated content was rare, and the overarching concern was about videos.”

However, things have changed quite rapidly, and the manipulated content policy is not enough for today’s advancements. Meta’s ‘Made with AI’ label will newly appear on video, audio, and images that have been AI-generated and thus manipulated.

The appearance of such a label on created content will be based on the detection of “industry-shared signals of AI images” and self-disclosures by people who have uploaded a content piece created by or with AI. Images generated by AI will specifically include the label ‘Imagined with AI.’

The company’s new labeling policy will be implemented starting in May this year. While labeling AI-generated content, Meta doesn’t plan to remove it entirely — well, unless the content violates already existing policies. The company will also stop removing content based on the previous manipulated video policy in July 2024.