Article 50 transparency duties for chatbots, synthetic content, emotion recognition and deepfakes; harmonised standards and presumption of conformity; who enforces the Act; the right to explanation; and the fines in Articles 99 and 101.
The provider's duty to design chatbots and other AI that interacts directly with people so that those people know they are dealing with an AI system, the obviousness and law enforcement exceptions, and timing.
The provider's duty to mark AI-generated audio, image, video and text in a machine-readable, detectable way, the technical standard it must meet, the December 2026 grace period for existing systems, and the code of practice that supports it.
What deployers must disclose: telling people exposed to emotion recognition or biometric categorisation, labeling deepfakes, and disclosing AI-generated text published on matters of public interest, with the artistic and editorial exceptions.
How providers can presume conformity through harmonised standards, common specifications and cybersecurity certification, the 2025 standardisation request, and the first AI Act standard cited in the Official Journal in October 2026.
Who enforces the AI Act: the AI Office and AI Board, national notifying and market surveillance authorities, the status of national designations, and the AI Office's expanded supervisory and fining powers since the 2026 amendment.
The right of people affected by a deployer's decision based on an Annex III high-risk system to a clear explanation of the AI's role and the main elements of the decision, its limits, and the related notice and documentation duties.
The three tiers of maximum administrative fines for operators and notified bodies, the lower caps for SMEs and small mid-caps, the factors authorities weigh, and Member State penalty rules, including fines on EU institutions under Article 100.
The Commission's power, from 2 August 2026, to fine general-purpose AI model providers up to 3% of worldwide turnover or EUR 15 million, the infringements covered, and the procedural safeguards.
Tell us what you're working on, whether it's a deployment, an audit, a security test or a cyber range. You'll speak with an engineer who can help you scope it.