Skip to main content
Audact
See the offer

Deadline · 2 August 2026

Article 50: what changed on 2 August 2026, and what did not.

In force since 2 August 2026

Last reviewed: 5 October 2026

Since 2 August 2026, Article 50 of the EU AI Act requires that a person is told they are interacting with an AI system, unless that is obvious from the circumstances, and that synthetic audio, image, video and text are marked in a machine-readable way. It is a transparency duty. It does not prohibit deploying AI, there is no licence to obtain, and it carries no certificate and no harmonised standard, so no vendor can be certified for it.

What the law requires

A person interacting with an AI system must be informed of that, unless it is obvious from the circumstances and the context of use.
Regulation (EU) 2024/1689, Art. 50(1)
Providers of systems that generate synthetic audio, image, video or text must mark the output in a machine-readable format, detectable as artificially generated.
Regulation (EU) 2024/1689, Art. 50(2)
The information must be given at the latest at the first interaction or exposure, clearly and distinguishably.
Regulation (EU) 2024/1689, Art. 50(5)
Breaching Article 50 carries a ceiling of €15 million or 3% of worldwide annual turnover. For SMEs and start-ups the ceiling is the lower of the two, not the higher.
Regulation (EU) 2024/1689, Art. 99(4) and Art. 99(6); Commission guidelines C(2026)5054 of 20 July 2026, point (152)
A full provenance chain is not mandatory. The Commission guidelines say so in as many words.
Commission guidelines C(2026)5054 of 20 July 2026, point (73)
Article 50(2) only applies from 2 December 2026 to systems that were placed on the market before 2 August 2026.
Digital Omnibus (EU) 2026/1744, in force 27 July 2026, inserting Art. 111(4)

What Audact does about it

The disclosure is spoken inside the call flow, at the position the policy for that country requires (Germany active · Netherlands in progress · Belgium, Austria queued), not written down in a policy document.
In controlled testing
A receipt is visible in the product for governed calls. The evidence record holds the approved disclosure text, its language, and the moment its audio finished being sent.
Working today
Policy differs per jurisdiction, because the national layer on top of Article 50 differs.
Germany: national policy layer active. Netherlands: in development. Belgium and Austria: planned.
Machine-readable marking of synthetic audio under Art. 50(2). It would be part of a later release; no date has been established and it is not in the product today.
Planned

The difference that gets missed

A transparency duty is not a ban.

The most common sentence written about 2 August 2026 is that every EU deployment became illegal that day. It did not. Article 50 says what has to be told and what has to be marked. It does not say who may deploy, and it creates no approval, no register and no certificate. Anyone offering an “AI Act certification” for a voice agent is offering something the Regulation does not contain.

The second thing that did not change is the burden of proof. Article 50 requires the disclosure; it does not require you to prove afterwards that it happened. Whether you can show it becomes relevant only when someone asks: a client, a supervisory authority, a court. That gap between meeting a duty and being able to demonstrate that you met it is the whole reason a receipt exists.

Who carries it

Provider or deployer: the distinction that decides your obligations.

The Act splits duties between the provider of an AI system and the deployer who puts it to use. Telling a person they are speaking to an AI falls on the provider under Article 50(1), and so does marking synthetic output under Article 50(2).

This matters most for agencies. An agency that puts its own name on a voice agent and sells it to an end client is, on most readings, the provider of that system rather than a deployer of somebody else's. That is a different set of duties and a different exposure, and it is worth settling which side of the line a given setup falls on before the first call rather than after it.

Note

Information, not legal advice.

This page sets out the rules as we read them, with the article numbers so you can check us. It is not legal advice, and your own position depends on facts this page cannot know. Take advice from qualified counsel before you rely on it.

Questions

Does Article 50 ban AI phone calls?
No. It is a transparency duty. It requires that the person is told and that synthetic output is marked. Deploying an AI voice agent in the EU is lawful; doing it without the disclosure is what creates exposure.
What is the maximum fine for a small company?
For SMEs and start-ups the ceiling is the lower of €15 million or 3% of worldwide annual turnover. That rule is in the Regulation itself, in Article 99(6) read with Article 99(4); point (152) of the Commission guidelines of 20 July 2026 restates it. The formulation “whichever is higher” circulates widely because that is what Article 99(4) says for undertakings in general — Article 99(6) reverses it for this group, and reading only Article 99(4) is how the wrong version keeps being repeated.
Do I need a full provenance chain on every piece of audio?
Not as a matter of law. Point (73) of the Commission guidelines of 20 July 2026 says a full provenance chain is not mandatory. What Article 50(2) requires is that synthetic output is marked in a machine-readable way.
I am switching a system on now. Does the December date buy me time?
No. Article 111(4) reaches only systems that were already on the market before 2 August 2026; for those, Article 50(2) starts on 2 December 2026. Anything switched on after 2 August 2026 is covered from the first call. That includes every Audact deployment: nothing of ours was on the market before that date, so no Audact pilot is grandfathered and none of them waits for December. Article 50(1), the duty to tell the person, was not postponed for anyone.

Article 50 inside the call, not in a document

Audact places the disclosure in the call flow and records that it happened. See how the policy engine decides what is said, in which country, at which point in the conversation.

See Governed Voice AI

Related