Sometimes — but there’s no rule that every AI-made image or post needs a “made with AI” tag. Two narrower duties apply: AI tools must invisibly mark what they generate, and you must disclose deepfakes and AI-written public-interest articles.
The Act separates (1) machine-readable marking — an invisible signal baked into AI output so software can detect it, which is the job of the tool provider (Midjourney, ChatGPT and the like), not you — from (2) disclosure to people, i.e. telling your audience, which only applies in specific cases.
This is the big misconception. The Act does not require every AI-assisted image, product photo or marketing post to carry a label. Ordinary AI-assisted marketing content is largely outside it, and minor touch-ups and standard edits are specifically carved out.
Deepfakes: realistic AI-generated or -manipulated images, audio or video of real-looking people or events must be disclosed as artificial (with a lighter touch for obviously artistic or satirical work). AI-written public-interest text: text published to inform the public on matters of public interest (news, health, politics, consumer info) must be disclosed — unless a person has genuinely reviewed it and holds editorial responsibility. A spell-check or a rubber stamp doesn’t count.
The disclosure duties apply from 2 August 2026. The invisible machine-readable marking has until 2 December 2026 for tools already on the market before August.
Not sure where you stand? The free checker tells you in two minutes which duties apply to you — no signup, no data collected.
Run the free 2-minute check →Other common questions