The date businesses had been watching with mixed feelings arrived on Sunday: August 2, when the European Commission’s AI Office began enforcing the AI Act and its transparency rules became binding. Most of the attention around the law has focused on its high-risk requirements, which the bloc recently pushed back to 2027 and 2028. But the obligations that took effect this week are arguably broader, because they touch any AI system that talks to people or generates content.
Four duties are now enforceable. Chatbots and interactive assistants must inform users that they are dealing with a machine rather than a human. Deepfakes, meaning images, video, or audio that have been generated or edited with AI, must be labeled as such. AI-generated or edited material must also carry machine-readable markings that other software can recognize. And deployments of emotion recognition or biometric categorization must notify the people exposed to the system. Separate duties cover AI-written text used to inform the public, which the Act subjects to disclosure when it deals with matters of public interest.
The obligations fall on both sides of the market: providers, who build the systems, and deployers, who put them to use. They apply to any organization serving the EU market regardless of where the company is headquartered, and open-source systems are not exempt.
The relief that many in the industry felt came from the AI Omnibus, the simplification package the EU agreed in May and which entered into force on July 27. That deal deferred the high-risk rules to December 2027 for areas such as biometrics, education, and employment, and to August 2028 for AI embedded in products like lifts and toys. The deferral for transparency is much narrower: only the machine-readable marking requirement for generative systems already on the market before August 2 is pushed back, to December 2, 2026. Chatbot disclosure, deepfake labeling, and the emotion-recognition notice all apply from day one, and any new system launched after August 2 must mark its output from the start.
Enforcement carries real teeth. Non-compliance with the transparency duties can draw fines of up to 15 million euros or 3 percent of global annual turnover, whichever is higher. The Commission published guidance on the rules in July and has released a list of more than 180 organizations that signed its voluntary code covering transparency of AI-generated content, the instrument that operationalizes the labeling duties.
For businesses, the practical message is that the widely reported delay concerned high-risk AI, not the everyday transparency rules. A chatbot that does not say it is a chatbot, or a synthetic video posted without a label, is now a compliance risk in Europe as of this week.
Sources: Celebrating the AI Act delay? The EU AI Act’s chatbot and content rules apply this week (TechRadar, Aug 3, 2026); Commission starts enforcing AI Act rules and new transparency requirements on 2 August (European Commission, Jul 31, 2026); EU AI Act Article 50: Transparency Rules for Businesses (GDPR Local, 2026); The EU AI Act’s Transparency Rules: A Practical Guide to Article 50 (EU Artificial Intelligence Act, May 14, 2026); One Month to Go: EU AI Act Transparency Compliance (McCann FitzGerald, Jul 2, 2026)