What changes this week
Discord starts sorting every account into one of three groups this week, adult, teen or unconfirmed, using signals it already holds rather than a document a user has to hand over. More than 90 percent of accounts will be classified without anyone doing anything, based on how long the account has existed, which servers and games it touches, and how active it has been. Adults see no change. Teens keep messaging and voice calls but lose access to age-restricted servers and sensitive images, and their profiles stay visible only to friends. Unconfirmed accounts get the same restrictions as teens until the system, or the user, resolves the ambiguity.
| Group | Messaging and calls | Age-restricted content | Profile visibility |
|---|---|---|---|
| Adult | Unchanged | Full access | Unchanged |
| Teen | Unchanged | Blocked | Friends only |
| Unconfirmed | Unchanged | Blocked | Friends only |
The breach this redesign is built to avoid
Discord's first attempt at age checks, announced in February 2026, asked users for a government ID or a video selfie through a third-party vendor, and it collapsed within weeks. That same vendor had already exposed government ID photos and verification selfies belonging to about 70,000 users in a breach a few months earlier, which meant the company was asking for the exact category of data that had just leaked. Discord retired that vendor, delayed the global rollout to the second half of 2026, and rebuilt the system to confirm most people without ever asking for a document at all. Anyone who still needs to confirm an age can now do it through a Stripe card check, an app-store age flag, or an on-device face scan Discord says never leaves the phone.
The gap the rollout does not cover
The account-signal model that spares most users an upload does not run in the UK or in Australia, where Discord's own description of the rollout says the estimation tool is currently unavailable, even though age checks already launched in both countries under local law. That gap lines up with a position UK regulator Ofcom put in writing in 2026: estimating a user's age from behavioral patterns, rather than verifying it against a document or a payment method, does not meet the evidence bar the Online Safety Act expects of a platform Discord's size. A method built to work almost everywhere without asking anyone for anything stops at the one border where a regulator has already said that kind of guess is not enough.
What this sets up for every other platform
Discord built its account-signal model to avoid repeating February's breach, not to satisfy a regulator, and it is now running two different products in two different legal environments as a result, an inference-based one everywhere the law allows it, and a document-based one wherever it does not. Any platform betting on behavioral signals to avoid asking users for an ID will meet the same UK-shaped hole, and the EU's own age-verification guidance under the Digital Services Act is moving toward demanding hard evidence too. The lesson for an owner building age or identity checks into a product is that the cheapest method to build is rarely the one every market will let you ship.
Read next: Loot Boxes for Minors Are Now Illegal in the EU's Own Draft Law | Grok's Designer Is The Defendant, Not The Platform



