Conversation

Notices

1. Embed this notice
   Daniel Supernault (dansup@mastodon.social)'s status on Saturday, 01-Aug-2026 18:35:10 JST Daniel Supernault

   That also raises an interesting point, will fediverse projects need to begrudgingly adopt AI to protect their communities, or will they experience mod/admin burnout with manual approval, and how will this impact the fediverse a year from now?

   I may have to rewrite my #FediCon talk a bit, because this is not theoretical anymore.

   We need to do something about fediverse spam, impersonation and AI agents.

   • Embed this notice
     Taran Rampersad (knowprose@mastodon.social)'s status on Saturday, 01-Aug-2026 18:45:37 JST Taran Rampersad

     in reply to

     @dansup that is an interesting problem. I don't know that there are any good options.

   • Embed this notice
     Daniel Supernault (dansup@mastodon.social)'s status on Saturday, 01-Aug-2026 18:50:21 JST Daniel Supernault

     in reply to

     • 〽️ɪɢᴜᴇʟ

     @bitbraindev yeah, we do offer curated onboarding in Pixelfed and Loops for this, but at some point, the admins need to manually review each one and that can become overwhelming, especially if targeted by attackers.

   • Embed this notice
     〽️ɪɢᴜᴇʟ (bitbraindev@mastodon.gamedev.place)'s status on Saturday, 01-Aug-2026 18:50:22 JST 〽️ɪɢᴜᴇʟ

     in reply to

     @dansup I guess instances need to adopt strict manual approvals for signups. Before even a join request is in, Mastodon should offer verification tools (e.g. domain verification like it works now) and then based on domains/verification quality an assessment can be made. Ironically, to automate this, an LLM could be utilised.