Article 50: the AI in your hiring funnel has to say it is AI

Edition 15. A reading time of about 4 minutes.

The hook

Most of the EU AI Act coverage you have read is about two things: whether your hiring tool is high-risk, and how large the fines can get. Both matter, and both have timetables that are still being argued over. Article 50 is the part that is easy to miss precisely because it is quieter. It says that when a person deals directly with an AI system, they have to be told. A screening chatbot, an automated interviewer, a voice assistant that calls candidates back, each of these has to make clear it is software, not a person. And unlike the high-risk timetable, this duty is not in the part of the Act that may slip. It applies from 2 August 2026, and the Commission reaffirmed that date this month.

What this means

Article 50 sets transparency duties that are separate from, and lighter than, the high-risk regime (AI Act, Article 50). The one that touches hiring most directly is in paragraph 1. The text reads: "Providers shall ensure that AI systems intended to interact directly with natural persons are designed and developed in such a way that the natural persons concerned are informed that they are interacting with an AI system, unless this is obvious from the point of view of a natural person who is reasonably well-informed, observant and circumspect, taking into account the circumstances and the context of use" (Article 50(1)). In plain terms: if a candidate is talking to your AI, the candidate has to know it is an AI.

Two features of that sentence change how you should read it. The first is who it binds. The duty is written onto the provider, the company that builds and supplies the system, not on you as the employer using it. If you license a screening chatbot, the obligation to design the disclosure in sits with the vendor. That is an honest distinction, and it cuts both ways: the statutory design duty is not directly yours, but the practical risk of a candidate reaching your AI without being told lands in your process, your brand and your hiring record. You are the one the candidate is dealing with.

The second is the exception. The disclosure is not required when the AI is "obvious" to a reasonably well-informed, observant and circumspect person. That carve-out is narrower than it looks. A pop-up plainly labelled as a bot may qualify. An AI that mimics a human recruiter over email, or a synthetic voice that sounds like a person on a call, is the exact case the rule is built for. As a rough guide: the more human-like the system, the less you can lean on "obvious".

Paragraph 5 sets the timing of the disclosure itself. The information "shall be provided to the natural persons concerned in a clear and distinguishable manner at the latest at the time of the first interaction or exposure" (Article 50(5)). So it has to be upfront, not buried in a privacy policy a candidate finds later.

One point worth holding onto: Article 50 applies whether or not your tool is high-risk. The high-risk classification under Annex III is a separate, heavier track. A simple screening chatbot that never reaches the high-risk threshold can still owe this transparency duty. It is a floor, and it is broad.

Now the date, which is where this becomes timely. Article 50 sits in Chapter IV of the Act, which is not carved out of the general application date, so it applies from 2 August 2026 (AI Act, Article 113). What makes that worth flagging now is what is happening around it. The Commission's Digital Omnibus package, which would push some high-risk deadlines later, remains provisional and is not yet in the Official Journal. The transparency duties are not part of that delay. On 10 June 2026 the Commission published a voluntary Code of Practice on marking and labelling AI-generated content and tied it explicitly to the "transparency obligations that will apply from 2 August 2026", noting that "users must also be informed when they are interacting with an interactive AI system, such as a chatbot" (European Commission press release, 10 June 2026). So while the high-risk calendar is still in motion, the transparency date is fixed and was just reaffirmed.

What to do with this

You do not need a lawyer to take the first steps. Three moves, and they are mostly about knowing what you run and what your vendor built.

1. List every candidate-facing AI touchpoint. Screening chatbots, automated interviewers, voice screeners, AI that drafts and sends replies. For each, check one thing: does it tell the candidate it is an AI, clearly, at the first interaction. If you are not sure, you have found the gap.
2. Make disclosure a vendor question, in writing. The design duty is the provider's, so ask them how the tool informs candidates and when. Get the answer in the contract, not in a sales call. You are the deployer; you cannot redesign their system, but you can refuse to run one that hides what it is.
3. Do not rely on "obvious" for human-like tools. If your AI sounds or reads like a person, treat the disclosure as required and make it explicit. The exception is for cases where no reasonable person could be confused, and a convincing AI recruiter is not one of them.

Read this way, Article 50 is one of the simpler duties in the Act, and one of the few with a date that is not moving. The teams that get ahead of it are not buying new software. They are checking that the AI already in their funnel introduces itself, and getting that confirmed in writing before August.

This is educational material and a starting point, not legal advice. The EU AI Act is still being amended, and parts of the high-risk timetable are among the provisions proposed for change under the Commission's Digital Omnibus package, which is not yet final law. The Article 50 transparency obligations discussed here are not part of that proposed delay and apply from 2 August 2026 under the Act as published, but how any duty applies depends on the specific facts and on national implementing rules. For your situation, consult a qualified lawyer. Signato is not a law firm and does not certify compliance.

Not sure whether the AI in your hiring funnel introduces itself? Take the free AI Hiring Risk Self-Assessment to see where you are exposed, then see the same evidence-first discipline applied to what your AI sends out: the free local trial of the Signato chamber blocks a draft that pairs the wrong person with the wrong subject before it leaves, no card, nothing leaves your machine.

Every claim here is traced to the primary regulation and checked by a person before it goes out. We tell you what is settled, what is still open, and what to do next. Know someone who uses AI in hiring in the EU? Forward this to them.

Sources (primary, read 2026-06-15):

- AI Act, Article 50 (Transparency obligations for providers and deployers of certain AI systems), Article 50(1) requires providers to design AI systems intended to interact directly with natural persons so that those persons are informed they are interacting with an AI system, unless this is obvious to a reasonably well-informed, observant and circumspect person; Article 50(5) requires the information to be provided in a clear and distinguishable manner at the latest at the time of the first interaction or exposure. - AI Act, Article 113 (Entry into force and application), general application from 2 August 2026; Article 50 sits in Chapter IV, which is not within the carve-outs of Article 113(a), (b) or (c), so it follows the general date. - European Commission press release, 10 June 2026, "Commission publishes Code of Practice on marking and labelling AI-generated content", a voluntary code tied to the "transparency obligations that will apply from 2 August 2026", stating that users "must also be informed when they are interacting with an interactive AI system, such as a chatbot". Confirms the transparency date was not moved by the provisional Digital Omnibus package. - Canonical text: Regulation (EU) 2024/1689, EUR-Lex CELEX:32024R1689.