Urgent.News

What's breaking now, across thousands of outlets.

AI

Discord deploys machine learning to classify user ages

Discord is rolling out a machine-learning system that estimates whether users are teenagers or adults from account and behavioural signals, allowing more than 90% of its users to avoid submitting identification or completing a facial age check. The messaging platform began deploying the system globally this week as part of a broader age-assurance programme designed to place accounts into adult,…

Discord has introduced an AI-powered system to estimate users' ages based on account and behavioral signals, allowing the majority of its users to bypass identification or facial age checks. This age-assurance program categorizes accounts into adult, teen, or unclassified groups, with teens defined as users aged 13 to 17 and adults as those 18 and older. The minimum permitted age on Discord remains 13.

The system analyzes various account attributes, such as account age, subscription history, email or phone verification status, activity patterns, device information, and social connections. However, it does not analyze message content, uploaded media, usernames, or profile biographies.

Central to the model is Discord's Entity-Relationship Embeddings (DERE), which transforms user, server, and game relationships into numerical representations. These social-graph embeddings prove to be the most predictive features, as people of similar ages tend to form clusters through friendships, communities, and gaming activities.

Discord employs XGBoost, a machine-learning technique that combines these embeddings with account metadata, engagement measures, community context, and system information. The company has excluded potentially biased attributes like race, gender, and ethnicity from the model's inputs.

Instead of forcing every account into an age category, the system generates a probability of an account belonging to an adult, with scores above a certain threshold resulting in an adult classification and scores below a lower threshold producing a teen classification. Accounts falling between these thresholds remain unclassified and receive additional safety protections until the user confirms their age.

The model's scores are aggregated over a longer period to ensure stability and reduce manipulation risks. Discord has not disclosed the exact thresholds but asserts that the model meets or surpasses the effectiveness of existing age-assurance methods in jurisdictions with age-related requirements.

Users classified as adults retain their normal experience, while teen accounts gain access to messaging friends, joining voice calls, and participating in non-age-restricted servers but face stricter safety settings, including restrictions on age-gated content and spaces. Users can view their assigned status through account settings. If the model cannot classify a user confidently or if they dispute an assignment, they can use alternative age-assurance methods, which vary by region and device.

For document-based checks, outside providers handle the process, and Discord receives only the resulting age result, not the person's identity. The model was trained on age-group labels generated through age-assurance deployments in 2025 and 2026, where users had confirmed their age via facial estimation or government identification.

This age-assurance approach follows a shift in Discord's rollout plans after a February delay, which involved adding verification choices, enhancing transparency, and publishing technical documentation. As age-assurance requirements expand globally, Discord has already implemented solutions in places like Texas and Brazil.

Written by urgent.news from Arabian Post's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at thearabianpost.com →

More in AI

Angie Nixon Proposes Building Data Center on Mar-a-Lago

The Florida Senate candidate told Newsweek that AI data center dividend checks should be an option to compensate Americans.

  • Florida Democratic Senate hopeful Angie Nixon proposes building AI data center at Mar-a-Lago.
  • Nixon calls for moratorium on new data centers to study environmental and resource impacts.
  • Nixon supports stricter AI regulations while remaining open to technology.

AI Safety Requires Eyes Inside the Lab | Opinion

"In a race toward systems whose controllability has not been demonstrated, being first...may simply mean losing control first."

  • AI safety demands unfiltered access to AI systems' inner workings.
  • Public oversight lacks for frontier AI compared to regulated industries.
  • Governments should pilot Anthropic's lab access for AI model supervision.

More from Saturday 26 September →