We name no specific product in the body of the site — these are the primary sources behind the patterns we describe. Marks like 1 on the main page link back to entries here.
Teen-by-default settings rolled out globally; users may be prompted to complete facial age estimation or upload a government ID to unlock age-restricted content and certain settings.
discord.com/press-releases/discord-launches-teen-by-default-settings-globally →An internal model uses account-level signals — tenure, payment method on file, servers joined, activity patterns — to predict age group. For users the model is unsure about, ID or face-scan verification is required.
discord.com/blog/getting-global-age-assurance-right →In late 2025, attackers compromised a third-party customer-support vendor used for age-related appeals, exposing roughly 70,000 users' government ID photos and selfies. A clear case of "the data was only leakable because it had been collected."
techcrunch.com → Discord age-verification rollout eff.org → Discord pushes mandatory age verificationAge-verification requirements are already in force in the UK, Australia, and Brazil, with Europe and multiple US states close behind — the reason platforms with user accounts are retrofitting ID-collection pipelines in the first place.
proton.me → Discord global age verificationThe opus target, per-client tx/rx bandwidth, CPU, RSS, edge latency, and matrix delivery shown under the Low / Med / High / Music buttons are pulled from real whisper-bench runs of the group_voice_synthetic scenario (8 peers, 2 active senders, 15s steady-state, debug build) recorded in the project's Echo instance. Low / Medium / High are Echo runs voice-profile-debug-{low,medium,high}-20260417-10044·; Music is Echo run voice-profile-debug-music-20260419-202914, captured for this page and re-ingested against the same scenario. MOS-LQO (ViSQOL) scoring is not yet wired into Echo and is called out as such on the landing.