EU will mandate labels on authentic-looking AI content starting August 2
The EU is mandating that companies stick a label on content that was AI-generated if it is attempting to look authentic, according to a report by The Guardian and others. These rules, which were devised under the region's pre-existing AI Act, go into effect on August 2.
This applies to generated images, audio and text, as long as they were designed to look real. This content will get slapped with a digital watermark. The rules do not apply to personal content, so custom group chat jokes are still safe. Also, there's an exemption for "evidently artistic" satirical and fictional works.
The August 2 deadline applies to new AI systems on the EU market. Pre-existing AI systems get four extra months to comply.
"It is a matter not only of customer protection, it's also a matter of democracy protection," EU parliament member Sergey Lagodinsky, who helped negotiate the landmark AI Act, said. "This information is something which we need to preserve our democracy and the authenticity of facts online."
Failure to comply could result in some serious fines, up to three percent of a company's total gross revenue. The EU has created its own black-and-white labels that anyone can use, but organizations can also make their own.
An AI policy lead with the Computer and Communications Industry Association told The Guardian that consumers may end up surprised where these labels show up. Boniface de Champris suggested that most AI-generated content on social media is already labeled, but consumers may be shocked to see them "in areas where AI is already used at scale but people don't know it — advertising, film, publishing."
It is true that large social media companies do typically have policies in place to label realistic AI-generated images, but the results aren't always foolproof. TikTok says it requires creators to label AI-generated content, but all kinds of content slips through the cracks. The latest trend is short videos of AI-generated doctors doling out dubious health advice.
Google has its SynthID tool, which it says has been used to label more than 100 billion images and, gasp, 60,000 years' worth of audio content. Meta has instituted policies to help people identify AI-generated photo-realistic images. This also includes various labels.
