What the Commission proposed
On 11 November 2026 the European Commission adopted its Digital Fairness Act proposal together with a Simplification Directive. Per the law firm Osborne Clarke's analysis, the text would prohibit designing online interfaces in ways "likely to deceive or manipulate consumers, or otherwise materially distort their behaviour." It would also blacklist ten dark patterns as unfair in all circumstances, and the Commission could update that list.
The items Osborne Clarke names include false availability or popularity claims, persistent scarcity messaging, defaults a consumer must reject, inferred consent to pay for extras, buried price or renewal information, fake re-engagement notifications, and making it hard to refuse or withdraw a choice. The proposal also requires a clearly labelled termination function for subscriptions, notices when free trials end, and renewal reminders. It adds rules on addictive design and personalised pricing. Traders would have 12 months to prepare once the regulation becomes law.
The strongest case for the rules
The case for acting is real. Dark patterns exploit predictable cognitive biases at scale, and existing consumer law has struggled to reach them. In its 2022 coordinated sweep, EU consumer authorities found manipulative practices on 148 of 399 online shops screened, per the Commission's January 2023 release. That was after years of the Unfair Commercial Practices Directive and its guidance being on the books. Case-by-case enforcement of a general fairness standard is slow, and honest firms that refuse manipulative tactics lose ground to those that use them. A bright-line list is the kind of rule that can change that.
Where the blacklist works
A per-se ban on specific, well-understood tactics is proportionate regulation at its best. Fake countdown timers, pre-selected add-ons and cancellation flows built as mazes have little legitimate defence. A list gives compliance teams a checklist, and it gives regulators a cheap enforcement route. Nobody needs to litigate whether a fake "only 2 left" banner was misleading in context. It is also a speech-neutral rule. It regulates conduct in a transaction flow, not what anyone says or reads, so it sits comfortably with an open-internet, pro-speech position.
The subscription provisions deserve similar credit. As Tech Policy Press' analysis of the proposal describes it, Article 23 requires a termination function labelled "terminate the contract here" or an unambiguous equivalent, and Articles 21-22 require renewal reminders with a free cancellation right if the reminder is missed. This is narrow, testable and cheap for compliant firms. Anyone who has been unable to cancel a subscription knows the problem is real.
Where it risks overreach
Three features deserve scrutiny.
First, the general ban on interfaces that "materially distort behaviour." Almost all design influences behaviour. A prominent button, a recommended plan and a well-timed prompt all do. A standard this open-ended invites uncertainty for startups that cannot afford a regulatory-affairs department. The blacklist is the part of the proposal that already does the work, so the general clause should be tied to evidence of consumer detriment, not to design intent alone.
Second, the movable list. The ability to update the blacklist is useful, since tactics evolve. Tech Policy Press' authors welcome delegated acts but want the power to cover the whole blacklist. The trade-off is that a per-se ban is only certain if the list is stable. Adding items by delegated act, with no co-legislator vote on each, shifts decisions about what counts as unlawful from Parliament and Council to the Commission. If the power is kept, additions should require an impact assessment, public consultation and a long lead time before they bite.
Third, penalties and compliance scale. Osborne Clarke reports fines of up to 6% of global annual turnover for breach. That is a heavy deterrent when the boundary of the general ban is fuzzy. Firms facing it will over-comply, strip out harmless personalisation and nudges, and ship worse products. Safe harbours for good-faith designs that follow published guidance would reduce that risk without weakening the blacklist.
Addictive design and personalised pricing: restraint is a feature
The proposal's treatment of these two areas is more restrained than the headlines suggest. For addictive design, Tech Policy Press says Article 17 requires tools to disable features such as autoplay, unprompted notifications and streak incentives. Osborne Clarke describes it as an opt-out requirement, not a blanket ban. Personalised pricing, per Osborne Clarke, is barred only where it is reasonably likely to harm a consumer and the trader knows with reasonable certainty that the consumer is a child or vulnerable. It is not a general ban on personalised prices.
That targeting is the right instinct. Dynamic and personalised pricing can lower prices for some buyers, and a blanket ban would have restricted legitimate commerce to catch a narrow harm. Critics who want a harder ban on addictive features should note that opt-out controls are easier to square with adult autonomy than prohibitions on design choices users may like.
Why it matters beyond Brussels
EU consumer rules tend to become global design defaults, because firms prefer one interface to many. A blacklist the size of ten patterns is easy to implement worldwide. An elastic general clause is not, and it may push firms toward the most conservative design everywhere. Other regulators, including India's, are already treating wallet-refund defaults and similar flows as potential dark patterns in consumer cases. A clear EU list would give them a template. A vague one would give them a precedent for ambiguity.
What to watch
The proposal now goes to Parliament and Council, and amendments are likely. The things to watch are whether the general ban gains an evidence-of-detriment threshold, whether list updates get procedural safeguards, and whether guidance and safe harbours arrive before the 12-month clock starts. The blacklist is good policy. The open-ended parts are where the legislative work should go.