Deceptive patterns

Deceptive patterns are interface designs built to work against the person using them: a cancel flow that goes in circles, a pre-ticked box that adds insurance, a "No thanks, I don't like saving money" decline button. The site deceptive.design describes them as features of apps, websites and AI systems that "stop you doing what you want, or steer you into harmful decisions you would not have made deliberately."

Where the term comes from

UX designer Harry Brignull coined the term "dark patterns" in 2010, when he set up darkpatterns.org to name and catalog manipulative design (deceptive.design). The site has since moved to deceptive.design, and now uses the name "deceptive patterns." Brignull's vocabulary has been picked up by regulators: the site notes it has been referenced in the EU Digital Services Act and the California Privacy Rights Act.

In September 2022 the US Federal Trade Commission published a staff report, Bringing Dark Patterns to Light, which credits Brignull with the term and uses it for "design practices that trick or manipulate users into making choices they would not otherwise have made and that may cause harm." The FTC grouped the tactics into four areas: misleading consumers and disguising ads, making it difficult to cancel subscriptions, burying key terms and junk fees, and tricking consumers into sharing data.

Common types

deceptive.design lists types including:

• Hard to cancel (also called "roach motel"): easy to sign up, very hard to leave.
• Sneaking and hidden costs: extra items or fees appear late in checkout.
• Preselection: options that benefit the business are ticked by default.
• Confirmshaming: guilt-tripping copy on the decline option.
• Fake urgency and fake scarcity: countdown timers and "only 2 left" messages that are not true.
• Trick wording and visual interference: confusing double negatives, or a faded "decline" next to a bright "accept."
• Nagging: repeated prompts until the user gives in.

Why it matters

They can work in the short term. The FTC report describes a study where dark patterns doubled the share of people who signed up for a dubious identity theft protection service compared with a neutral interface. But they cost trust, generate complaints and chargebacks, and increasingly carry legal risk.

How to avoid them

Do:

• Make cancelling as easy as signing up.
• Give accept and decline equal visual weight and plain wording.
• Show the full price, including fees, from the start.
• Leave optional add-ons and data sharing unticked by default.

Don't:

• Use countdowns or stock warnings that are not real.
• Shame people for saying no.
• Ask again and again after someone has declined.

In AI products

AI products have their own versions. Consent screens that bundle "use my chats to train models" into the terms, with the opt-out buried in settings, are a form of preselection and obstruction. Upsell prompts that interrupt a task mid-answer, or limit messages without saying when they reset, borrow from nagging and fake urgency. A chatbot that keeps a cancellation request in a loop of retention offers is the old hard-to-cancel pattern in a new interface. Clear data consent choices and honest usage limits are the alternative.