What's the new EU rule requiring AI content to be labeled, and does it affect me outside Europe?

Seeing news about the EU making AI labeling mandatory. What actually has to be labeled, and does this apply if I'm not in the EU?

Starting August 2, 2026, Article 50 of the EU's AI Act makes it legally mandatory (previously more of a voluntary/platform-by-platform practice) to disclose when content is AI-generated, if that content could otherwise be mistaken for authentic. This covers two main things: deepfakes and other realistic synthetic images/audio/video need a visible warning, and AI-written text published on matters of public interest needs disclosure too — while a technical, machine-readable marking is also required for synthetic media more broadly, so it can be detected by other systems (like platform moderation tools). There are meaningful exemptions: content that a human has genuinely reviewed and taken editorial responsibility for is exempt, as is basic assistive use like spellcheck, grammar correction, translation, or general writing help — the rule targets content presented as authentic, not any use of AI tools in the process. On whether it affects you outside the EU: yes, potentially — the AI Act applies based on where the content reaches users, not where the company or creator is based, so if your AI-generated content is directed at or accessible to people in the EU, the labeling requirement can apply to you regardless of your own location.