compares Threads to Mastodon, and finds the toxicity levels to be roughly the same.
(I'll point out that "Toxicity in the Decentralized Web and the Potential for Model Sharing" study is similarly Pleroma only, 729 Pleroma instances from 2020, most of which are no longer online)
@moubliezpas
This study: https://arxiv.org/html/2502.17926v2
compares Threads to Mastodon, and finds the toxicity levels to be roughly the same.
(I'll point out that "Toxicity in the Decentralized Web and the Potential for Model Sharing" study is similarly Pleroma only, 729 Pleroma instances from 2020, most of which are no longer online)
@moubliezpas The reference for "With toxic or abusive content being common in the Fediverse" is a single study (https://ojs.aaai.org/index.php/ICWSM/article/view/31293/33453) that looked specifically at Pleroma only, not the Fediverse.
Pleroma itself has its own baggage, and the study's own data is skewed in large part by one overwhelmingly toxic domain that is no longer active.
This is in no way a reasonable source for "toxic or abusive content being common in the Fediverse"
Also possibly useful: https://about.iftas.org/2025/03/27/content-classification-system-post-mortem/
@Jamie some recommended reading here for hosting ActivityPub services: https://github.com/FediFence/fedifence/blob/main/LegalRegulatory.md
Cloudflare has a free CSAM filter: https://developers.cloudflare.com/cache/reference/csam-scanning/
IFTAS is working on an opt-in CSAM scanner for service providers, follow this account to be notified
Lemmy moderators should fill out this needs assessment: https://cryptpad.fr/form/#/2/form/view/thnEBypiNlR6qklaQNmWAkoxxeEEJdElpzM7h2ZIwXA/