What Took Effect on August 2
<cite index="5-1">The European Union began enforcing key provisions of its Artificial Intelligence (AI) Act on August 2, requiring companies to disclose when people are interacting with AI and to label certain AI-generated or manipulated content.</cite>
<cite index="8-2,8-3">Article 50 of the EU AI Act requires providers to clearly inform people when they are directly interacting with an AI system unless that fact is obvious from the circumstances, and providers must also make covered AI-generated or manipulated content detectable through machine-readable markings.</cite>
<cite index="7-1">Deepfakes must be visibly disclosed, while AI-generated or manipulated text published to inform the public must be labelled when it has not undergone human review or editorial control.</cite> <cite index="7-7">People must also be informed when they are exposed to emotion-recognition or biometric-categorisation systems.</cite>
<cite index="2-15,2-16,2-17,2-18">AI that interacts directly with people — chatbots, virtual assistants, autonomous agents — must tell the user it is an AI the moment they first encounter it; fine print buried in terms of service will not suffice. The disclosure must be immediate, accessible, and clear. There are two exceptions: cases where it is obvious to anyone that they are dealing with an AI, and systems used for law-enforcement purposes.</cite>
Enforcement Architecture
<cite index="1-3">According to the European Commission's AI Act Service Desk, the AI Office and national authorities formally gained the power to enforce the rules on prohibited AI practices, transparency, and General-Purpose AI (GPAI) models from August 2, rather than simply monitoring compliance.</cite>
<cite index="11-7,11-8">On August 2, 2026, the Commission's enforcement powers in respect of GPAI model providers came into force. While the obligations of GPAI model providers came into force on August 2, 2025, providers were given an adjustment period of one year before the Commission could start exercising its supervision and enforcement powers against them.</cite>
<cite index="5-4">Enforcement is divided among the European Commission's AI Office, national regulators, and the European Data Protection Supervisor.</cite> <cite index="7-14,7-15">National market-surveillance authorities retain important responsibilities, while the EU AI Office has powers in relation to general-purpose models and certain systems built upon them. Cross-border cases may test how quickly these bodies exchange information and agree which authority should lead.</cite>
Penalty Structure
<cite index="35-10">EU AI Act fines run in three tiers under Article 99: up to €35 million or 7% of total worldwide annual turnover for prohibited AI practices, €15 million or 3% for most other breaches, and €7.5 million or 1% for supplying incorrect information.</cite> <cite index="36-8">Fines are calculated as the higher of the fixed euro amount or the revenue percentage for large organisations, and the lower of the two for small and medium-sized enterprises and startups — a critical proportionality mechanism.</cite>
The High-Risk Track: A Separate Timeline
The August 2 enforcement date does not mark a single compliance cliff for the AI Act as a whole. <cite index="23-3">The Digital Omnibus postpones the high-risk obligations for Annex III AI systems from August 2, 2026 to December 2, 2027, and the obligations for high-risk AI in regulated products under Annex I to August 2, 2028, but it leaves the Article 50 transparency rules and the Article 4 AI literacy duty exactly where they were.</cite>
<cite index="24-1">This legislative proposal, dubbed the 'AI Omnibus,' was adopted on November 19, 2025; a political agreement was reached on May 7, 2026, and it entered into force on July 27, 2026.</cite>
<cite index="18-4">Annex III covers the larger category for most enterprises: AI used in recruitment and employee management, creditworthiness assessment, biometric identification and categorization, education and vocational training access, migration and border control, law enforcement, and administration of justice.</cite> <cite index="21-6">That original August 2026 date was practically challenging, especially since the harmonised standards that providers need in order to demonstrate conformity were not going to be ready in time.</cite>
Code of Practice and Global Reach
<cite index="5-9">The Commission said more than 180 organizations have signed a voluntary Code of Practice designed to help companies apply the transparency rules for AI-generated content.</cite> <cite index="14-12">For signatories, the Commission will focus enforcement activities on monitoring adherence to the Code and may take commitments to the Code into account as mitigating factors when fixing the amount of fines; non-signatories are expected to demonstrate compliance via other adequate means.</cite>
<cite index="8-11,8-12">The practical consequence extends beyond European technology companies. American developers, platforms, retailers, and service providers offering AI products in the European market may need to adjust user notices and content provenance practices.</cite> <cite index="8-12">Generative AI systems placed on the market before August 2 may receive until December 2026 to satisfy the machine-readable marking requirement.</cite>