What the Digital Omnibus is and what changed
The European Union's AI Act came into force in 2024 with a phased schedule of obligations. In November 2025, the European Commission proposed simplifying it as part of a package called the Digital Omnibus. According to the law firm Cooley, the final text was agreed on June 29, 2026, was published in the Official Journal as Regulation (EU) 2026/1744 and entered into force on July 27, 2026, as Hunton also confirms.
The main postponements are:
- Stand-alone high-risk systems (for example, AI for hiring or in education): moved from August 2, 2026, to December 2, 2027.
- AI built into regulated products (toys, medical devices, machinery): from August 2, 2027, to August 2, 2028.
- National regulatory sandboxes (sandboxes): until August 2027.
The Omnibus also extended SME benefits to mid-sized companies and softened the staff AI literacy obligation, which the Commission and member states must now support.
What was not postponed: telling people they are talking to an AI
Article 50 of the law sets transparency obligations that have applied since August 2, 2026, and the Omnibus did not move them. The one that matters most to anyone serving customers is Article 50(1): if an AI system interacts directly with people, such as a customer service chatbot or a voice assistant, the person must know they are talking to a machine.
The Commission's guidelines on Article 50, published on July 20, 2026, and summarized by the law firm McCann FitzGerald, clarify how to comply:
- At the latest at the first interaction. The notice must be built into the system's design, not hidden in the terms and conditions.
- Any format. It can be text, audio or visual. For a voice agent, a spoken notice at the start of the call fits the guidelines.
- Answer truthfully. The system must disclose that it is an AI if the user asks or could be confused.
- Narrow exception. It is only exempt when it is obvious to a reasonably well-informed person that they are dealing with an AI; the guidelines call for applying this exception restrictively.
Fines for breaching the transparency rules can reach €15 million or 3% of worldwide turnover, whichever is higher.
What comes in December 2026
The next milestone is December 2, 2026. On that day the transition period ends for generative AI systems that were already on the market before August to mark their synthetic content (audio, image, video or text) in a machine-readable way, according to the Commission and the guidelines. To support that marking, the Commission published a voluntary code of practice in June 2026.
On that same day, new prohibitions added by the Omnibus take effect: AI systems designed to generate non-consensual intimate images and child sexual abuse material.
In short, the calendar looks like this: transparency toward users from August 2026, content marking for older systems and new prohibitions from December 2026, high-risk systems from December 2027 and AI in regulated products from August 2028.
For a voice agent this means two things: say from the very first moment that it is an AI and, if it generates synthetic audio, have the technology provider mark it according to the rules.
What it means for businesses in Colombia and Latin America
The EU AI Act can reach companies outside Europe when their systems are offered in the Union or their outputs are used there. But even if your company only serves customers in Colombia, the European law is a good best-practice handbook:
- Say it is an AI from the greeting. A short sentence at the start of the call or chat builds trust and prevents complaints.
- Always offer the option to talk to a person. It is the best way to avoid frustration.
- If you serve customers in Europe, check with your legal advisor that your chatbot or voice agent complies with Article 50.
- Protect the data. In Colombia, recordings and conversations are personal data protected by the Ley 1581 de 2012 (Colombia's data protection law).
At 19 80 TIC Cloud we set up AI voice agents and 24/7 chat that introduce themselves as AI and hand the conversation over to a human advisor when the customer needs it.
