EU requires labels on realistic AI content starting August 2
From August 2, 2026, the European Union will require AI-made images, videos, audio, and text that look real to be clearly labeled unless they are users' personal content or evidently artistic, satirical, or fictional works.
This new rule is part of the European Union's AI Act and is meant to fight misinformation and keep things fair in society.
New AI tools need to follow these rules right away; older ones have until December.
Companies face fines up to €15 million
Companies that don't label could get hit with fines up to €15 million or 3% of their global revenue. There are exceptions for obvious art, satire, or fiction.
Google has already watermarked over 100 billion images with SynthID, and TikTok has labeled billions of AI posts.
Regulators say they want transparency, but the tech industry also warns against overwhelming users with too many labels, so finding the right balance is key.