Created with AI? The EU orders companies to include a label that makes it clear

The goal? Ensure that Europeans know immediately whether the online content they see is real or fake.

The comprehensive EU law on artificial intelligence will be implemented in stages. Starting August 2, companies must ensure that their chatbots and other models make it clear to users that they are AI toolsor that the images or texts created with it carry a label indicating this.

This can be achieved by integrating watermarks and other indicators that allow AI-generated content to be easily detected. Companies face high fines if they do not comply with these provisions.

“Generative AI makes it possible to create disinformation on an unprecedented scale, tailored to specific audiences and spread with remarkable speed,” said an EU official.

Since AI “is making it increasingly difficult for us all to distinguish what is real from what is synthetic,” EU rules seek to “preserve citizens’ ability to trust what they see, hear and read”he added.

Of particular concern to the EU are “deepfakes” or content known as deepfakes: texts, images, videos and sounds that appear real, but are generated or manipulated by AI.

These new European rules will affect content created for professional purposes, and the EU insists that people using AI in a purely personal capacity will not be affected.

Texts intended to inform the public on topics of general interest must also be tagged if they are created using AI without human editorial supervision.

Existing artificial intelligence systems have until December 2 to adapt to the new rules, and there are exemptions for “artistic, creative, satirical and fictional” works.

The EU has come under fire for placing more demands on companies and setting rules that will ultimately force labeling of all content, given the widespread and rapidly growing use of AI across Europe.

“We’ve heard that it’s going to be very, very difficult to implement. But I think we often hear this with respect to compliance requirements. And yet the world keeps turning and we find a way to figure these things out,” Ashley Casovan of the International Association of Privacy Professionals told AFP.

Tech giants around the world have already started applying their own labels in anticipation of the standards.

For example, TikTok has required content creators to label AI-generated images, audio and videos for several years, and claims that more than 3 billion content items are already flagged using detection tools and technology.

Similarly, Meta has implemented an “AI Insights” flag on Instagram and Facebook for posts that use this technology.

Google, for its part, signed the EU code of conduct on transparency in artificial intelligence and says it is working with other companies such as Nvidia, OpenAI and Apple on digital labeling tools.

However, Google’s Karen Massin warned of “regulatory complexity” that could backfire and “risk confusing the people these rules are intended to help.”

“If online content is inundated with overlapping AI labels and legal notices, it will be harder for people to get the clear context they need,” he said.