Discord will publish public-channel messages to search engines from mid-December
Servers with at least 1,000 members that anyone can join get a public web page on November 16. Messages follow in mid-December, without usernames or avatars.
Admins can pick which channels participate or switch them all off, in Server Settings under Web Presence, and members get a single toggle in User Settings under Data and Privacy. Public channels are on by default, except member settings default to off in the EU, UK, Canada, Brazil, Australia and South Korea.
The thread worried less about Google than about who can be linked back.
What Discord is changing

Per Discord's announcement, the change covers only Discoverable Servers: those with at least 1,000 members and open joining. Private servers, DMs and group DMs are not affected. Every such server gets a public web page on November 16, 2026, with a default built from its profile if the admins set nothing. Automated channel summaries may appear from the same date, and from mid-December public-channel messages may appear on web pages and in search results without usernames or avatars. Messages sent before mid-December are not included.
Admin controls roll out from October 27. Members have one switch covering all Discoverable Servers. Turning it off stops future messages appearing, but messages already shown stay unless deleted. Teen users are excluded.
The risk that isn't training data
A commenter argued the real danger is deanonymization:
"People posted in a 'private' community under nicknames."
The server administrator enables discoverability for everyone, he wrote, so the whole history of members who never opted in becomes discoverable and linked to their accounts, and the same nickname on GitHub or X points to the same person.
Checking the opt-out
A commenter who had just checked said an account-level opt-out exists and that theirs was already off, apparently inheriting an earlier data opt-out. Kagi's founder and CEO noted that Google appears nowhere in the post and searching the web is broader than Google. Another is tentatively hopeful, calling it a disgrace how many communities moved from the open internet into Discord.
Members who joined a big server years ago should check the User Settings toggle before mid-December, since removal after the fact does not recall what is already indexed.
Get breakdowns like this every day
The top Hacker News stories, unpacked — by email, Telegram, or RSS.
A crawler pulled 145.5 GB from 3D-model sites in 34 hours, including 54.8 GB of adult content
The user agent was "trellis2-data-sourcing/1.0", with a Roblox-linked contact address. Open3DLab says it ignored robots.txt and the sites' terms.