Discord’s new AI decides if you’re an adult, and restricts you if it’s not sure

Discord is rolling out an in-house machine learning model designed to verify the age of its users and automatically restrict experiences for younger users.

In a blog post, the social messaging platform said that the model assigns users to an age group based on behavioural patterns and signals associated with their account.

Discord, which has 200 million global active monthly users, did not specify exactly what those signals are. However, it did confirm the model does not use message content.

Starting in March, the model will give users who cannot be verified a “teen-appropriate experience” experience by default, locking age-restricted servers, sensitive content and certain settings behind age verification.

Users who want to prove they are adults can record a short video selfie for facial age estimation, or submit a form of ID to a vendor partner.

Discord said the selfie is processed entirely on the user’s device and never stored. Identity documents, it said, are deleted shortly after confirming a user’s age.

The announcement initially prompted backlash, with users interpreting the changes as a mandatory face scan for everyone.

Discord added a clarification to the post within 24 hours stating it was “not requiring everyone to complete a face scan or upload an ID to use Discord.”

It said the vast majority of users would never be asked to confirm their age at all.

Explicit checks would only be needed where the background model could not confirm someone as an adult and that person wanted access to restricted content.

Drew Benvie, head of social media consultancy Battenhall, told the BBC that while “the sentiment behind creating a safer community for all social media users is a positive move,” putting the new system into place across millions of communities could be “fraught with issues.”

“Discord could lose users if its implementation of age verification backfires, but it could equally attract more new users who will be drawn to its new standards for online safety by design,” he said.

Privacy campaigners have previously warned that age verification methods such as face scans and ID uploads could pose a risk to users’ privacy.

Discord faced criticism in October after official ID photos of around 70,000 users were potentially leaked after a firm which helped it verify ages was hacked.

It said that the vendors handling its new age assurance process were not involved in that breach.

Previous Article

Research finds AI tools don't reduce work, they intensify it

Next Article

Claude Opus 4.6 arrives with a memory that can handle an entire book

View Comments (1)

Comments are closed.