Blog / Regulation

AI Act Article 50: what your agents must disclose since 2 August

This is the AI Act obligation the Omnibus did not postpone. It looks simple; it still requires knowing where, when and to whom each agent speaks.

By Vincent Dorange, , 2 min read

2 Aug 2026

the transparency obligations of Article 50 apply

Since 2 August 2026, the transparency obligations of Article 50 of the AI Act have applied 1. The Digital Omnibus adopted in July did not postpone them; it only gave systems already on the market until 2 December 2026 to mark generated content 2. For a company whose agents talk to customers, candidates or suppliers, this is the most immediate obligation in the Act.

What the text says

Paragraph 1 is addressed to providers: a system intended to interact directly with natural persons must be designed so that they are informed they are interacting with an AI, unless this is obvious from the point of view of a reasonably well-informed, observant and circumspect person, taking into account the circumstances. Paragraph 2 requires providers of systems that generate text, images, audio or video to mark those outputs in a machine-readable format.

Paragraphs 3 and 4 are addressed to deployers. A deployer of an emotion recognition or biometric categorisation system informs the people exposed to it. A deployer that publishes a deep fake discloses it, as does one that publishes AI-generated text to inform the public on matters of public interest, unless the text has undergone human review under someone's editorial responsibility. Paragraph 5 sets the timing: the information is given in a clear and distinguishable manner at the latest at the first interaction or exposure.

What it means for an agent

An agent that answers the phone, writes emails to customers or negotiates with a supplier falls under paragraph 1. The obviousness exception is best read narrowly: a well-written email sent in an adviser's name does not make it obvious that an AI wrote it. The safest course is to state the nature of the sender in the first message.

Above all, the obligation assumes the company knows which agents talk to whom, which is rarely the case where every department has plugged in its own tools. An up-to-date AI inventory that records, for each agent, the channels it uses and the people it reaches is the first compliance document. The Agentic Commerce Framework® adds a question: among those exchanges, which ones commit the company through a promise, a price or a refusal? Those are the ones that must be traceable in an audit trail.

Sources

Links to the official texts and original publications.

  1. 1Regulation (EU) 2024/1689 (AI Act), Article 50, Official Journal of the European Union, L series, 12 July 2024.
  2. 2Regulation (EU) 2026/1744 of 8 July 2026 (Digital Omnibus on AI), Official Journal of the European Union, L series, 24 July 2026.

Glossary terms used in this article

AI inventoryAI audit trail

Vincent Dorange, creator of the Agentic Commerce Framework®

Also worth reading

95%

of organisations surveyed faced at least one agent-related incident in 12 months

Governance

95% of organisations have already had an AI agent incident, says Gartner

25

organisations interviewed; none lets its agents act with unrestricted autonomy

Governance

AI agents at work: what the OECD learned from 25 organisations

4 levers

transaction caps, budgets, approved counterparties and escalation rules, per Circle

Agentic economy

Agentic economy: agents that spend within delegated permissions

All articles