Holos Discover - Federation & Moderation Policy

shield Our moderation stance

Holos Discover is a consent-based Fediverse search engine. Because we index content at scale, we take a proactive approach to moderation: we rely on trusted third-party denylists, we react quickly to incidents, and we default to removal when in doubt. We never publish posts of our own — our only outbound activities are Follow and Undo Follow.

For the user-facing consent model (how we decide whom to index), see How it works. For user privacy rights and opt-out, see our Privacy Policy. This page focuses on moderation of instances.

gpp_bad Blocked instances policy

We maintain a public list of blocked instances. Blocks come from four sources:

1. IFTAS DNI list (Do Not Interact)

We import the IFTAS DNI list in full. This list is curated by IFTAS through its SW-ISAC programme and covers content that poses legal or safety risk to operators — primarily CSAM (child sexual abuse material), TVEC (terrorism and violent extremism) and NCII (non-consensual intimate imagery). Entries on this list are obfuscated on our public stats page to avoid advertising illicit domains.

2. IFTAS AUD list (Abandoned & Unmanaged)

We also import the IFTAS AUD list, which catalogues abandoned or unmanaged instances that are a common source of spam, LLM-generated content, and coordinated disinformation campaigns.

3. Manual blocks

The operator may block additional instances for the following reasons:

blockSpam, harassment or coordinated inauthentic behaviour observed in indexed content
link_offBridges and republishing services that import content without the original author's consent (e.g. Bluesky/Nostr bridges that force indexable=true, RSS-to-AP bots)
warningInstances whose moderation posture is incompatible with ours (no-moderation policies, hosting of banned words, etc.)

4. Automated detection

We run scheduled checks that can result in automatic blocks:

hubdetect-bridges: samples posts to identify instances that bridge Bluesky content (at://did: patterns) without user consent
do_not_disturbcheck-suspensions: detects instances that have suspended our actor and blocks them in return (double-checked via signed + unsigned requests to avoid false positives)

report Illegal content reporting

If you find indexed content that is illegal in your jurisdiction — in particular CSAM, TVEC or NCII — please contact the instance administrator immediately. Reports can come from anyone (users, instance admins, law enforcement, IFTAS).

scheduleAcknowledgement within 24 hours, action as soon as the report is confirmed
delete_foreverConfirmed illegal content is hard-deleted from our index, and the source instance is evaluated for blocking
gpp_goodSerious or systemic cases are reported to IFTAS SW-ISAC and, where appropriate, to the competent authorities

gavel Appeal / unblock process

If you are an instance administrator and believe your instance has been blocked in error, you can request a review:

mail_outlineContact the instance administrator by email, stating your domain and the reason you believe the block is unjustified
ruleWe re-evaluate the block against the source (IFTAS DNI/AUD, manual or automated) that triggered it
check_circleIf the block is lifted, your instance is un-blocked and content indexing can resume on consenting users

Note: blocks originating from the IFTAS DNI list cannot be lifted on our side — you must engage directly with IFTAS for removal from their denylist.

visibility Transparency

The full list of blocked instances is published on our public stats page, grouped by source (IFTAS DNI, IFTAS AUD, manual). Domains labelled CSAM, TVEC or NCII are obfuscated (TLD + SHA-256 hash prefix) — an operator can verify whether their own domain is on the list by hashing it locally, without us publicly advertising illicit domains.

mail Contact

For moderation questions, illegal content reports, or unblock requests, contact the instance administrator. We welcome dialogue with other instance admins, IFTAS, and participating platforms.