EU’s AI Act Fines of Up to €35 Million Spur Compliance Race as New Deadlines Emerge
27.06.2026 - 05:35:23 | boerse-global.de
The European Union’s artificial-intelligence rulebook is coming into sharper focus, and businesses face escalating penalties if they fall behind. With potential fines climbing to €35 million or 7 percent of global annual turnover for banned practices, companies across Europe are scrambling to build compliance frameworks before the AI Act’s key dates arrive.
High-risk systems get extra time — but not unlimited
The European Parliament approved adjustments to the regulation with 423 votes in favour and 57 against. Under the revised timetable, standalone high-risk AI systems listed in Annex III of the regulation enjoy a transition period that now ends on 2 December 2027. Providers whose AI acts as a safety component inside regulated products have even longer: they must comply by 2 August 2028.
Yet the core calendar holds firm. Full application of the AI Act begins on 2 August 2026. Since February 2025, bans on unacceptable-risk AI — such as biometric categorisation or social scoring — have already been in effect.
Transparency triggers in August 2026
Companies dealing directly with individuals face a pivotal deadline on 2 August 2026. Any AI system that interacts with people — for instance, chatbots used in human resources — must be clearly labelled as such. Emotion recognition at the workplace and deepfakes come under strict disclosure rules. Synthetic content must also carry machine-readable markings, with existing systems allowed until December 2026 to adapt.
A practical requirement with teeth is the obligation to ensure AI literacy under Article 4. Companies must train employees on how to work with AI. Although the rule formally took effect in early 2025, it becomes enforceable only from August 2026 — meaning violations can then trigger fines of up to €15 million or 3 percent of global annual turnover.
Germany sets up its own oversight
The German legislature moved in mid-June, passing the KI-Management-Gesetz (KI-MIG). The Federal Network Agency becomes the central market-surveillance authority, while specialised agencies handle sector-specific controls.
Beyond Germany, some EU states are taking a tougher line. Italy has declared purely AI-based automated decisions in employment relationships void. In North Rhine-Westphalia, policymakers are pushing for a complete ban on social scoring and automated performance monitoring at the workplace.
Worker fears and adoption figures drive urgency
Market data underscores the compliance race. In 2025, roughly 19.8 percent of German companies were using AI. The political target is 75 percent by 2030. At the same time, 41 percent of German professionals fear losing their job to the technology.
Nearly 48 percent of companies already deploy AI agents — often without sufficient controls. For compliance departments, standardised audit processes and zero-trust architectures are becoming essential tools to avoid heavy penalties and reputational damage.
