EU AI Act UX Requirements: What Changes in Your Product
The EU AI Act affects product UX in four concrete ways: users must be told when they interact with AI or see AI-generated content, high-risk systems need interfaces that let humans oversee and override them, deployers need understandable instructions, and manipulative techniques that cause significant harm are banned outright. Most of these land on screens, flows and microcopy.
This article translates the regulation into design decisions. It complements our broader AI agent compliance guide and focuses on what designers, product managers and UX writers should change. It is not legal advice, and parts of the timeline may move, so verify the current status before you plan releases.
What is the timeline that matters for product teams?
The Act entered into force on 1 August 2024. The prohibitions and the AI literacy duty applied from 2 February 2025, obligations for general-purpose AI models from 2 August 2025, and most remaining obligations, including the Article 50 transparency rules, apply from 2 August 2026. Certain high-risk rules follow in 2027.
In late 2025 the European Commission proposed a digital omnibus package that may delay or simplify parts of the high-risk regime. Treat the dates above as the baseline, track the legislative process, and design for the stricter reading where the cost is low. Clear disclosure and good oversight are good UX anyway.
Which manipulative designs are now prohibited?
Article 5 bans AI systems that use subliminal, purposefully manipulative or deceptive techniques, or exploit vulnerabilities related to age, disability or social and economic situation, where this materially distorts behaviour and causes or is likely to cause significant harm. The threshold is high, but the direction is clear for designers.
In practice, review AI-driven persuasion features with extra care: personalised urgency messages, emotionally tuned retention flows, companions that discourage users from leaving, or agents that steer vulnerable users toward costly choices. Our article on dark patterns, AI and regulation covers the consumer-law rules that catch manipulative designs well below the AI Act threshold.
How should transparency look in the interface?
Article 50 requires that people are informed when they interact directly with an AI system, unless it is obvious from the context, and that certain AI-generated or manipulated content is disclosed. The information must be clear, distinguishable and provided at the latest at the first interaction or exposure, and it must meet accessibility requirements.
Design implications are concrete: a visible label at the start of a chat, not only in the terms; consistent markers on AI-generated images, audio or text where disclosure is required; and screen-reader-friendly wording. We go deeper in our piece on transparency obligations for chatbots and agents.
Want an expert opinion on your product?
Talk to a UX expert or AI Code Engineer — 30 minutes, free, no obligation.
Book a call for freeWhat does human oversight mean for high-risk interfaces?
For high-risk systems, Article 14 requires that natural persons can effectively oversee the system. That is a UX requirement in disguise. The people overseeing need to understand capabilities and limits, notice anomalies, avoid over-relying on output, interpret results correctly, and decide not to use or to override them, including a way to stop the system.
Translate this into interface features:
- Confidence or uncertainty indicators that are honest rather than decorative.
- Explanations of the main factors behind a recommendation, in the user's vocabulary.
- An override path that is as easy as accepting the suggestion.
- A clear stop or pause control for running processes.
- Friction against automation bias, for example asking reviewers to inspect evidence before approving in bulk.
Does AI literacy affect onboarding and training?
Article 4 requires providers and deployers to take measures to ensure a sufficient level of AI literacy among staff and others operating AI systems on their behalf. For internal tools this is partly an onboarding problem. In-product guidance, sensible defaults, short explanations of limitations and examples of good and bad use can carry part of the load that classroom training cannot.
As a rule of thumb, if a new employee could misuse your AI feature on their first day without noticing, your onboarding and microcopy need work, regardless of what the training records say.
Measure it: ask a few new users what the AI feature cannot do after onboarding. If they cannot name a single limitation, the guidance is not landing.
How do you turn the Act into a UX backlog?
A UX audit focused on these touchpoints is one of the fastest ways to find gaps, because it looks at the product as users and regulators will experience it.
- List every AI touchpoint in your product: chat, recommendations, generated content, automated decisions.
- Mark which ones involve direct interaction, generated content, or high-risk use cases.
- For each, define the required disclosure, oversight control and documentation.
- Write and test the microcopy with real users, including screen reader users.
- Add regression checks so disclosures do not disappear during redesigns.
- Have an independent reviewer audit the flows before the obligations apply to you.
Frequently asked questions
Do I need an AI label on every screen?
Not necessarily. The obligation is to inform people when they interact directly with an AI system, unless it is obvious, at the latest at first interaction. A clear label at the start of the conversation plus consistent markers where content disclosure applies is a common pattern.
Is personalisation banned under the AI Act?
No. The prohibition targets manipulative or exploitative techniques that materially distort behaviour and cause or are likely to cause significant harm. Ordinary personalisation is not covered, although consumer and data protection law still apply.
Who is responsible for the UX of oversight features?
The provider of a high-risk system must design it so that oversight is possible; the deployer must assign competent people to use those features. In practice both sides should test the oversight interface together.
How can we check our product before August 2026?
Run an inventory of AI touchpoints and commission a focused UX review of disclosures, consent and oversight flows. UXreviewer.io delivers such reviews with commented screenshots within 10 business days; see pricing.
Book a call for free
From UX audits and Agentic UX reviews to AI development, agentic coding workshops and publishing your AI-built app on the App Store and Google Play — we help you ship. 24/7 call support in English and German.