I checked out the EU’s new AI labeling rules, and there's a huge loophole hiding in plain sight

The big loophole in this legislation

The landmark EU AI Act clarifies that this labeling applies to the aforementioned content only if it's designed to look real.

This creates a conundrum, as what's "designed to look real" is subject to interpretation. Personal content is excluded from the law, as well as "evidently artistic" satirical and fictional works, making the rules even more vague.

The EU has created black-and-white labels anyone can use, but AI systems and companies can also choose to design their own, as long as the content is clearly marked.

Huge fines for violators

The EU has also drafted the fines for all companies found to be violating the new rule. These fines are up to €15m (£12.8) or 3% of a company’s worldwide global turnover, which in the case of companies like Apple, Google or Samsung, results in could potentially result in billions of dollars.

The August 2nd deadline applies to new AI systems, but all existing ones have four more months to comply. Once again, this law concerns companies that operate in Europe, so don't expect changes in the US and the rest of the world yet.

Companies already mark AI-generated content

That said, companies and big social networks and apps already have policies in place when it comes to AI-generated content.

Google even has a special SynthID tool, which, according to the company, has been used to label more than 100 billion images and 60,000 years' worth of audio content. That's huge and also shows how widespread AI-generated content has become.

However, despite these policies, AI images and videos constantly slip through the cracks and make it to people's feeds. With this new EU policy in place, companies will need to be more strict or face the financial repercussions.