EU AI Act: since 2 August, chatbots must say they are AI and deepfakes must be labelled
Archive · Issue of 7 August 2026

What happened
On 2 August 2026 the EU’s AI Office and national authorities started enforcing the AI Act, and a set of transparency rules began to apply. The European Commission confirmed the date in a press release on 31 July (European Commission). Chatbots must tell people they are talking to an AI. Deepfake images, video and audio must be labelled. Breaking the transparency rules can cost up to EUR 15 million or 3% of worldwide turnover, whichever is higher (Cooley).
Why it matters for your business
Most of the work falls on the companies that build AI tools, which must mark AI output in a machine-readable way. For generative AI tools already on the market, that duty has a grace period until 2 December 2026 (Cooley).
The duties for businesses that use AI did not get a grace period. If you publish a deepfake, such as an AI-generated video of a person, you must say so. If you put a chatbot in front of customers, it should be obvious they are talking to a machine.
The stricter rules for “high-risk” AI (used in hiring, credit and similar decisions) are being pushed back. In May, EU lawmakers agreed through the Digital Omnibus to move them to 2 December 2027, and to 2 August 2028 for AI built into regulated products (Gibson Dunn).
The Swiss angle
Being based in Zurich or Lugano doesn’t take you out of scope. The Act applies to anyone whose AI systems reach the EU market or whose AI outputs are used in the EU, wherever the company sits (Cooley). If you sell to customers in Germany, France, Italy or Austria, this applies to you.
At home, Switzerland’s Data Protection Act already covers AI. The Federal Data Protection Commissioner says people have a right to know whether they are talking to a machine, and that manipulated faces or voices must always be clearly marked (FDPIC). A Swiss AI law is still in preparation: the Federal Council plans a consultation draft by the end of 2026 (OFCOM).
What to do
Check every place a customer meets AI on your behalf: website chat, WhatsApp bot, AI images of people. Where it’s missing, add one plain line, such as “You’re chatting with our AI assistant. Ask for a person any time.”
Sources
- European Commission, press release, 31 July 2026: Stichtag 2.8.2026: Durchsetzung der Vorschriften des KI-Gesetzes und neuer Transparenzanforderungen (English summary: IP/26/1714)
- Cooley, 3 August 2026: EU AI Act: Transparency Obligations Take Effect 2 August 2026
- Gibson Dunn, 27 May 2026: EU AI Act Omnibus Agreement: Postponed High-Risk Deadlines
- FDPIC: AI and data protection
- OFCOM: Künstliche Intelligenz
Legal note: This article is general information, not legal advice. Laws change and every business is different, so always get advice from a qualified lawyer or your fiduciary (Treuhänder) before you act on it.
Get the free newsletter
Every Friday: three AI stories that matter for Swiss businesses, and what to do about them. Free, in English, German, French or Italian.