September 18, 2026 · 8 min read
AI Act: what changed for websites and apps in August 2026
Europe’s new transparency rules for certain AI systems are now being enforced. Here is what digital teams should review.
From August 2, 2026, the European Commission and national authorities began enforcing new AI Act transparency obligations. Not every AI feature carries the same risk or duties, but hiding automation is no longer a prudent product decision.
When a user is talking to AI
European guidance requires certain chatbots and interactive systems to clearly tell people they are interacting with AI, unless that is obvious from the context. The notice should arrive at the right moment and be understandable — not buried in a policy nobody reads.
Synthetic images, video and audio
Deepfakes and certain AI-generated or manipulated content require identification. The Commission also refers to machine-readable markings that make detection easier. Product design should accommodate this information without turning every screen into a legal warning.
A practical product checklist
- Inventory where AI generates, changes, recommends or decides.
- Separate editorial assistance from automated user interaction.
- Explain briefly when AI is present and which data it uses.
- Keep human review for sensitive or consequential decisions.
- Record vendors, versions, consent and ways to challenge a result.
Transparency is also design
Good implementation is not a frightening pop-up. It is a clear label, an accessible explanation and a human alternative where appropriate. Trust is part of the interface.
Important note
This article is general product and communication guidance, not legal advice. The correct treatment depends on the system, sector, data and company role; projects with meaningful risk should be reviewed by qualified legal professionals.
Like how we think? Let us talk.
