Since 2 August 2026, Article 50 of the European AI Act has required detectable marking for certain synthetic content and disclosure for deepfakes or AI-generated text intended to inform the public. This legal requirement is reshaping the web content production pipeline: design is no longer just about appearance; it must incorporate asset traceability from the concept stage.
AI marking and metadata: technical requirements for design teams
The European AI Act sets out different levels of obligation depending on the actor’s role (provider or deployer) and the type of content (image, video or text). For design teams, this means that every visual generated by a model must include machine-readable marking before publication.
An image file exported from a generation tool must retain its provenance metadata through to final display. Any removal, whether deliberate or accidental, during an optimization step (compression, WebP conversion or processing through a CDN) exposes the site deployer to risk.
For the interface, disclosure must be noticeable to users without requiring any additional action. A discreet icon buried in a footer is not enough for deepfakes. Creators who follow MYN Idee news regularly find explainers on how these regulations apply to design.
A common pitfall is using AI to produce a rough draft, retouching it manually, and then treating the result as human-created content. The boundary remains unclear, but the regulation also covers content that has been “significantly modified” by an AI system. Hybrid workflows therefore require clear internal documentation.

Web accessibility in 2026: moving from WCAG 2.1 to WCAG 2.2
The European Accessibility Act has applied to relevant services, including mobile apps, since 28 June 2025. It is no longer a best-practice recommendation. Initial proceedings show that its scope also covers apps, not just desktop websites.
WCAG 2.2 is gradually replacing WCAG 2.1 as the reference standard in the European framework. The new criteria added in this version directly affect designers’ work:
These new criteria challenge overly subtle custom focus styles, impose constraints on layouts where the structure of help varies significantly between sections, and affect the design of multi-step conversion flows.
An audit based solely on WCAG 2.1 will miss these criteria. Product teams that have not updated their acceptance checklists are working from an incomplete standard.
Practical impact on design systems
Focus tokens, form components and navigation patterns need to be reviewed. A button whose focus ring relies on a 1 px outline in the primary color against a white background is likely to fail criterion 2.4.11. Design systems that have remained unchanged since 2023 are accumulating measurable accessibility debt.
Remediation costs rise when these checks are carried out after development. Incorporating WCAG 2.2 criteria into Figma components or design tokens avoids treating accessibility as a post-delivery fix.
AI-assisted creation: redistributing creative work
An Adobe study published in June 2026 describes an increase in creative job postings in the United States between September 2025 and April 2026. AI skills are being spelled out more often in job listings. AI is shifting creative work rather than eliminating it.
Professionals interviewed for the study use AI for certain tasks (visual exploration, adapting formats and generating variations) while setting limits around originality and control over the final result. Creative judgment—the ability to assess what works and discard what does not—is becoming the differentiating skill.
In interface design, this redistribution is reflected in a shift in deliverables. Pixel-perfect mockups are giving way to systems of rules and constraints that guide the automated generation of layouts. A designer who understands token logic and adaptive constraints delivers more value than one who draws every screen manually.

Transparency and visual identity
The challenge for brands is not to hide their use of AI, but to maintain a recognizable identity despite tools that tend to homogenize results. Default-generated palettes, “safe” compositions suggested by models, illustrations with converging styles: the main risk of AI in design is uniformity, not a loss of quality.
Successful teams set strict stylistic constraints upfront: a fixed typography guide, a deliberately limited color palette and a proprietary composition grid. AI then operates within a framework that preserves the brand’s distinctiveness.
Distinctive web design: beyond surface-level trends
Annual trend lists (3D, brutalism, dark mode, oversized typography) are widely circulated. When thousands of sites adopt them at the same time, the result is precisely the opposite of what was intended: uniformity disguised as modernity.
A deliberate typographic choice using no more than two type families creates more personality than five stacked visual effects. Visual hierarchy, page rhythm and the behavior of scroll-triggered micro-interactions matter more than whatever graphic style is currently in vogue.
Far from stifling creativity, accessibility and AI transparency requirements act as generative frameworks. A well-designed focus ring contributes to an interface’s identity. An AI provenance label carefully integrated into the layout builds credibility rather than undermining it.
In 2026, the quality of online design is measured by its ability to meet three requirements at once: visual distinction, regulatory compliance and honesty about how content is produced. Teams that treat these three areas as a single brief, rather than as separate constraints, create the most coherent interfaces.



