The Digital Fairness Act: The Interface Patterns With a Limited Future

The Digital Fairness Act is targeted for the fourth quarter of 2026, which means it is a Commission proposal rather than a law and will not apply before roughly 2028 to 2030. That timeline makes it easy to ignore.

Ignoring it is a mistake for a specific reason. Unlike most regulation, this one targets interface patterns rather than data handling or documentation. Interface patterns get built into products, tested, optimised and then depended on commercially. If your conversion funnel currently relies on something the Act is aimed at, you want to know that while you still have time to find an alternative that works.

What it is aimed at

The Commission has been consistent about the target list across its consultations and announcements. Five categories:

Dark patterns. Interface designs that steer users towards choices they would not otherwise make, by exploiting how attention and defaults work rather than by providing information.

Addictive design. Mechanics deliberately built to maximise engagement time: infinite scroll, variable reward schedules, streaks, artificial scarcity of attention, notification patterns tuned for return visits.

Personalisation that exploits vulnerability. Targeting based on inferred emotional state, financial distress, addiction susceptibility or age.

Personalised pricing. Showing different prices to different users based on inferred willingness to pay.

Influencer marketing. Disclosure obligations and responsibility allocation between platforms, brands and creators.

The unifying thread is asymmetry. Existing consumer law assumes a reasonably informed consumer making a decision. These practices work by making the interface itself the decision-maker.

Why the existing rules are considered insufficient

Dark patterns are already partly covered. The Unfair Commercial Practices Directive prohibits misleading and aggressive practices. The Digital Services Act contains a dark patterns provision for online platforms. GDPR requires that consent be freely given. Consumer rights law requires clear pre-contractual information.

The Commission's fitness check concluded that this coverage is fragmented and hard to enforce. The DSA provision applies to platforms rather than to all traders. The UCPD requires showing that a practice materially distorts behaviour, which is arguable case by case. GDPR consent rules only reach consent-related patterns.

The Digital Fairness Act is the attempt to consolidate that into something enforceable against a specific list of practices.

The patterns most likely to be named

Nothing is settled, but the Commission's consultation documents and the academic and enforcement literature converge on a fairly stable list. If your product uses these, they are worth examining.

False urgency and scarcity. Countdown timers that reset, "only 2 left" claims that are not tied to inventory, "17 people are viewing this" figures that are generated rather than measured. This is the most commonly enforced pattern already and the most likely to be explicitly prohibited.

Confirmshaming. Decline options worded to make the user feel foolish or irresponsible. "No thanks, I don't want to save money."

Preselected options. Anything opted in by default that costs money, shares data or extends a commitment.

Asymmetric friction. Subscribe in one click, cancel in six steps through a retention flow. This is already being addressed separately in several jurisdictions and it is very likely to appear here.

Drip pricing. Revealing mandatory fees progressively through checkout so the headline price is not the payable price.

Nagging. Repeated prompts for a permission or upgrade after refusal. Notably, the Digital Omnibus already proposes a six month silence period after a consent refusal, which is the same principle arriving earlier.

Disguised advertising. Content that is commercial but formatted to look editorial or user-generated.

Obstruction of the free choice. Making the privacy-protective or cost-free option available but harder to find, which is the pattern most cookie banners still use.

The two that will cause the most argument

Addictive design

This is the hardest to legislate because the line between engaging and addictive is genuinely blurry, and because the mechanics involved are shared by products nobody wants to ban.

Infinite scroll is used by news sites and social platforms alike. Streaks are used by language learning apps and gambling apps. Variable rewards describe both a loot box and a search result. Any rule specific enough to prohibit the harmful case risks catching the benign one.

Expect this part to end up focused on minors, on specific mechanics with clear evidence of harm, and possibly on design defaults rather than prohibitions.

Personalised pricing

Currently legal in the EU with a disclosure obligation added by the Omnibus Directive. Making it unlawful, or unlawful in some cases, would be a significant change and it is strongly opposed by parts of the retail and travel sectors.

The likely landing zone is somewhere between prohibition and the current disclosure requirement: perhaps a prohibition where the personalisation uses vulnerability signals, or where it is based on protected characteristics or inferred financial distress.

If your pricing engine personalises, the useful question now is not whether it is legal but whether you could explain the basis of a specific price to a regulator. Many pricing models cannot, because the model is opaque even to its operators.

What is worth doing before there is a text

The answer is not to redesign your product against an unpublished proposal. It is to know where you stand.

Inventory the patterns you actually use. Walk your own funnels: signup, checkout, consent, subscription management, cancellation, upsell. Note every place where the interface is doing persuasive work rather than informational work. Most teams find things nobody deliberately decided to build.

Find the ones you depend on commercially. This is the important step. If a pattern is worth two percent of conversion and it appears on the likely prohibition list, that is a business risk with a three year fuse. If it is worth nothing measurable, remove it now and stop worrying about it.

Fix the asymmetries first. Cancel should be as easy as subscribe. Decline should be as easy as accept. Reject should be as prominent as agree. These are the clearest cases, they are already enforceable under existing law in several member states, and they are the least likely to survive any version of this Act.

Make prices honest at the first display. Drip pricing is on every list, is already restricted in several jurisdictions, and is a genuinely bad experience. Showing the payable price up front costs some measured conversion and reduces cart abandonment and complaints.

Check what your personalisation actually uses as inputs. Not what you think it uses. If a model has access to signals that correlate with vulnerability, financial distress or age, you have a problem waiting whether or not anyone intended it.

Watch the age-related provisions closely if you serve minors. Everything aimed at minors will land harder and earlier, and it intersects with age verification obligations arriving through other routes including the EU Digital Identity Wallet.

The argument for doing this anyway

There is a version of this article that treats the Digital Fairness Act as a threat. That is not quite right.

Dark patterns work in the short term and are measurable, which is why A/B testing produces them reliably. What A/B testing does not measure well is the second-order cost: refund rates, chargebacks, support load, complaint volume, churn from users who felt tricked, and the reputational drag that accumulates slowly and is never attributed to the specific pattern that caused it.

The companies that have removed the worst patterns generally report a small conversion loss and a larger reduction in downstream cost. That is not a universal result and it depends heavily on the business model, but it is common enough that the conventional wisdom deserves testing rather than assuming.

The regulatory direction here is unusually clear and has been for years. Every jurisdiction that has looked at this has moved the same way. Building a growth model on patterns that are being legislated against in every major market is a strategy with an expiry date.

Where this fits

The Digital Fairness Act sits alongside the Empowering Consumers Directive, which applies from September 2026 and already restricts a set of commercial practices, and the consent rules being reshaped by the Digital Omnibus. The full legislative queue is mapped in our EU digital law pipeline.

Getting help

We build and improve consumer-facing products for companies operating in Europe, including the work of removing patterns that are earning short-term conversion and long-term liability without breaking the funnel.

If you want a review of where your product sits against the likely prohibition list, or help redesigning a flow that currently depends on friction asymmetry, write to office@c9group.dev. More about our European work on the EU market entry page.

We are engineers and product people rather than lawyers. Whether a specific pattern is currently lawful is a question for your counsel, and the Act itself does not exist yet.