Meta to tag AI-generated images on Instagram and Facebook

‍

Good news! Meta will soon be rolling out a tool capable of automatically flagging AI-generated content on its social networks.

Meta will label images produced by generative Artificial Intelligence systems such as Midjourney or DALL-E, announced Nick Clegg, the company's President of Global Business, in an interview on ABC's " Good Morning America ".

The "tags", due to be rolled out over the coming months, will identify images generated by Artificial Intelligence and posted on Facebook, Instagram and Threads. "As the difference between human and synthetic content blurs, people want to know where the line is. We're developing this functionality now, and in the coming months we'll start applying tags in all the languages supported by each app. "says Meta's blog.

The need to develop common markers

Meta's teams are working with a number of industry partners to develop common technical standards (IPTC metadata, watermarks, etc.) that will enable the identification of AI-created elements. The ability to detect these signals will enable Meta, and more broadly all content providers, to inform Internet users about the origin of images.

In the absence of common standards, Meta already applies labels to images created by itsown tool, Imagine with Meta. To date, however, the company is unable toidentify AI-generated audio and video content from external platforms. Indeed, while Google, OpenAI, Microsoft, Adobe, Midjourney and Shutterstock are beginning to include tags in their image generators, they are not yet including them in their generative AI tools.

Meta to flag content created with the help of artificial intelligence

Pending common identification criteria, in the coming months Meta will be offering a new feature allowing users to indicate when they are sharing AI-generated video or audio so that the company can tag them. "We will require users to use this disclosure and tagging tool when they post organic content with photorealistic video or digitally created or edited realistic audio. Failure to comply with this new rule will result in sanctions. The tagging feature will be available on all platforms.

But that's not all. Given the volume and complexity of AI-generated content, Nick Clegg acknowledges that tags won't be a "perfect solution". "People and organizations who actively want to deceive people with AI-generated content will look for ways around the safeguards put in place to detect it. In our industry and in society at large, we'll have to keep looking for ways to stay one step ahead." To limit the risks of circumvention, the company has indicated that it is working on "classifiers" that will make it more difficult to remove or modify metatags.

‍______________________________________________________________

Keep up to date with the latest news on social networks by subscribing to our newsletter.

_________________________________________________________________

‍

Identify content generated by generative AI to combat aberrations

As Americans head to the polls in 2024, White House spokeswoman Karine Jean-Pierre welcomes the initiatives of social platforms to offer visibility to Internet users and combat misinformation: "As social media companies make their own independent content management decisions, we believe they have an important role to play in enforcing their own rules to prevent the spread of misinformation and intimate, non-consensual images of real people." Meta plans to tag AI-generated images in the coming months as "a number of important elections are taking place around the world", said Clegg.

‍