Discord Will Guess Your Age From Account Habits After a Vendor Breach Exposed User IDs

Discord will start reading your account habits to guess how old you are. Beginning Wednesday, the chat platform rolls out age assurance to every user, and for most people the check happens quietly in the background rather than through a selfie or an ID scan.

More than 90% of users will not have to verify anything, CTO Stanislav Vishnevskiy wrote in a blog post. Instead of asking for documents, Discord feeds existing account signals into a machine learning model: how long the account has existed, the device and activity data tied to it, and other usage patterns. The model sorts each account into adult, teen, or unconfirmed.

The company is blunt about how that works. "The way a typical 15-year-old uses Discord looks different in aggregate from how a typical 30-year-old uses Discord," it wrote in a separate technical post, and the model has learned those patterns. Discord says it does not estimate age from message content, profile data, or demographic attributes, and that it is reading data it already retains rather than collecting anything new.

If the model is not confident a user is an adult, there is still an off-ramp that avoids biometrics. That user can appeal by sharing a credit card or the age-group data attached to their Apple App Store or Google Play account, with no government ID or face scan required for that path.

The backlash this rollout is answering has a specific cause. Discord delayed these plans in February after users objected to biometric scans and ID checks, and their worry was not abstract. Last year a third-party vendor Discord worked with was breached, potentially exposing about 70,000 users' government ID photos and selfies used for identity verification. The new signal-based approach is partly an attempt to sort ages without amassing another pile of that kind of sensitive data.

For confirmed adults, nothing changes. Teen accounts get walled off from age-restricted servers and channels, message requests from non-friends land in a separate inbox, and teens see a warning when they accept a friend request from someone with no mutual connections.

Vishnevskiy framed the move as a response to age assurance rules that keep expanding, noting the company already launched checks in Brazil and Texas after the February delay. Governments worldwide are passing these mandates faster than platforms can build for them, which is why the pattern of regulators leaning on platforms to police their own users keeps recurring. The result is a familiar bind for the industry, one that runs alongside the broader fight over who gets to set the rules for AI and online platforms: comply with a patchwork of local laws, or risk the penalties.