Meta to Introduce ‘AI-Generated’ Labels for Media starting next month

M

Starting next month, Meta is set to introduce a new feature across Facebook, Instagram, and Threads: a label that reads “Made with AI” on images and videos identified as AI-generated.

Recently, Meta shared in a blog post an update on its policy regarding AI-generated content. Come May, a “Made with AI” label will be attached to various types of content found on its platforms.

The deployment of this new policy is expected to cover content on Instagram, Facebook, and Threads. Meta will start indicating more video, audio, and image content as AI-generated beginning next month.

Content will receive these labels based on user disclosures about the use of AI tools or when Meta identifies “industry standard AI image indicators.” However, specific details on Meta’s method for detecting AI-generated images were not disclosed.

After taking into account suggestions and feedback from Meta’s Oversight Board, the company is planning to revise the manipulated media policy established in 2020.

The previous policy mainly banned videos altered by AI tools to make it seem like a person is saying something they did not. Meta acknowledged that the current policy is “too narrow” and fails to address the broad spectrum of AI-generated content that has gained traction on the internet lately.

Meta mentioned in its blog post, “In the past few years, and especially within the last year, there has been a rise in various forms of realistic AI-generated content like audio and photos, with advancing technology.”

Addressing manipulations that depict individuals engaging in actions they did not partake in is deemed equally crucial according to the Board’s observation.

Meta underlines the importance of adding transparency and supplemental information to these AI-derived images through the introduction of “Made with AI” labels. The company argues for labels and context as a preferable method to tackle manipulated media without unduly impinging on freedom of expression.

In an earlier announcement, Meta disclosed its collaboration with industry collaborators to establish standard technical criteria for identifying AI content, encompassing video and audio, back in February.

Given the ongoing nearing of the 2024 U.S. election period and the emergence of deepfake images of politicians circulating online, Meta emphasizes the significance of enhancing transparency on AI-generated images at this point in time.