Fediverse Moderation and AI Accountability
A narrative about AI-assisted moderation practices and questionable content policies across Fediverse platforms, including instances of shadowbanning and censorship.
April 2026
Deceptichum banned from PieFed.World
Deceptichum was banned from the PieFed.World instance. The ban was imposed by [email protected] after a disagreement in the community.
Deceptichum banned from lemmy.ca
The user Deceptichum received an instance ban from lemmy.ca. The ban occurred after Deceptichum discussed the Fedihosting Foundation.
MrKaplan banned users and defederated instances from Anarchist.Nexus
MrKaplan banned multiple users and blocked entire communities. Additionally, MrKaplan defederated from Anarchist.Nexus over the users' anti-Zionist stance.
Lemmy.world announcement concerning defederation from dbzer0
Lemmy.world made an announcement regarding potential defederation from dbzer0. This action was the subject of discussion in the announcement thread.
Administrator Kaplan's defederation actions regarding anarchist.nexus
Administrator MrKaplan reportedly defederated anarchist.nexus privately without consultation with other administrators. Kaplan later stated in response to questions regarding network size.
Serinus accused of policing user comments on Lemmy.world
User @Serinus was accused of policing comment count limits within the Lemmy.world community. Serinus also allegedly claimed 'Misinformation' status for user comments.
Luminous5481 banned [email protected]
Luminous5481 banned [email protected] from the community agitprop memes. The stated reason for the ban was "zionism, genocide apologia," and the ban carried a 74-year expiry time.
Admin Emopunker removed user comment on Feddit.org
The user Andy_R stated that Admin Emopunker removed a comment posted by Andy_R on feddit.org. This sanction occurred after Andy_R was providing evidence to counter a user's hearsay regarding misinformation.
May 2026
Federated instances using LLM for comment evaluation
Some popular federated instances began using LLM-assisted moderation tooling. This tooling sends user comment history to OpenAI to analyze content for evidence of specific political ideology sentiment and related tropes.
Source ↗db0 was blocked by a moderator on piefed.social
A moderator on the piefed.social instance allegedly insulted users and subsequently blocked replies from db0. This action was followed by db0 locking the thread.
AI-assisted banning action recorded on piefed.world mod log
A moderator logged a ban action for user [email protected]. The log entry included an image and a link, which the author claims was provided as proof from the LLM output.
Source ↗Discussion concerning LLM-based moderation practices on Fediverse
Users discussed the use of LLM-based moderation, noting that sentiment analysis moderation is not new. Some users questioned the reliability and use of automated moderation as an excuse for action.
Source ↗Quokka.au instances shadowbanned from PieFed.Social Instance Chooser
Quokka discovered that its instance was excluded from the PieFed.Social Instance Chooser. The exclusion was stated by the developer to be for 'obvious reasons'. This impacted the visibility of Quokka.au to potential new users.
PieFed implemented 'trusted' modlog visibility feature
PieFed.social added a modlog feature where the reason for an action is only shown from designated 'trusted' instances. This limits the visibility of moderation history for instances outside the trusted list.
PieFed introduced instance silencing feature
A new update to PieFed allows for 'instance silencing.' Silenced instances' posts will not show in Popular or All feeds, and their communities will not appear in Starter packs.
PieFed.social allegedly shadowed Quokk.au from Discoverability
Quokk.au discovered it was shadowbanned from the PieFed.Social Instance Chooser. The shadowban reportedly prevented Quokk.au from being advertised on the flagship PieFed instance.
Rimu kicks user from all Matrix PieFed rooms
Rimu banned a user from all Matrix PieFed rooms. The banning occurred after the user attempted to promote non-discriminatory language. The action reportedly followed Rimu's post about stepping back from drama.
Source ↗Rimu banned multiple users on yepowertrippinbastards
The user [email protected] imposed mass instance bans on approximately 20 users. The user Quokka claims these bans were for speaking out against Rimu's behavior in the PieFed developer Matrix channels.
Source ↗Rimu issued mass instance bans
The user Rimu was accused of implementing mass instance bans. The bans allegedly targeted 20 users who spoke out against his behavior. The accusation implies the bans were related to instances of perceived disagreement.
Source ↗Rimu mass-banned users on PieFed developer Matrix channels
Rimu imposed mass instance bans against approximately 20 users. The basis for the sanction was allegedly 'harassment' by the recipients.
Source ↗Rimu Bans Users on PieFed
Rimu reportedly mass-banned 20 users on the PieFed developer Matrix channels. The bans were allegedly imposed under the charge of 'harassment'. The user initiating the post claimed the bans were retaliatory for speaking out against Rimu's behavior.
Source ↗