Content Policy

What is rejected at submission

Every piece of content posted to Peanut Forum — questions, merge proposals, host responses — is screened by an automated classifier before publication. Content is rejected when it falls into a small set of clearly illegal categories:

• Sexual content involving minors — blocked automatically at submission, hashed, and retained under the 90-day legal hold required by 18 USC § 2258A. Confirmed violations are reported to NCMEC's CyberTipline.
• Credible threats of violence against identifiable people
• Doxxing — posting another person's private contact information without consent
• Other content that is illegal under US federal law

Rejection is bounded by what is illegal, not what is unpopular. The platform takes no editorial position on political speech, controversial opinions, or content that some users find objectionable but is legal.

What stays visible — with tools

Content that is offensive but legal stays visible. Users have three tools that operate at the individual or research level, not as platform-level removal:

• Click-through badges — content that has been flagged by users or by an automated classifier as potentially offensive is collapsed behind a "tap to view" badge. The content is still in the leaderboard ranking; it's just not shown without an explicit click.
• Hide from my view — any user can hide a piece of content from their own view. This affects only that user; the content remains visible to others, and a placeholder shows where the hidden item would have appeared in their feed.
• Report — any user can flag content. Reports are logged in a public ledger queryable by researchers and authorities. Reports do not trigger removal or any automated action. The flag becomes part of the record; the platform does not act on it.

The public record

Peanut Forum is built around a transparency-first model. The flag system is a record, not a removal mechanism. Specifically:

• Every user-submitted report is recorded with the flagger's pseudonymous handle, the content it targets, and the reason category.
• Every automated classifier flag is recorded with the model name, category, and confidence score.
• Every hard block is recorded with the category, classifier verdict, and a SHA-256 hash of the rejected content (the content text itself is not retained, except where federal law requires retention for evidence purposes).
• Authorities and academic researchers may run any analysis they want against this record. We ask that interventions arising from such analysis be FOIA-discoverable so that the public can see what was acted on and why.

The research API exposes flag and block data through views rv_content_flags, rv_content_llm_flags, and rv_content_blocks.

Full event content is separately archived to Arweave, a permanent record outside our own database — see Event Archives.

Why we don't ban accounts

By design, Peanut Forum does not know who any user is — there are no real names, email addresses, or identifying details on file. Standing to participate is verified at the device level (hardware key, residency proof, organizational credential) and the platform receives only the outcome of that verification.

Because the platform cannot identify a person, it cannot meaningfully ban one. The closest analog is per-user content hiding: each user can decide what they don't want to see, and that decision is private to them.

This is a structural property, not a policy choice. There is no admin panel that could un-anonymize a user, because there is nothing in the database to un-anonymize.

Report a concern

Use this if you don't have a specific piece of content to flag from within the app or site — for example, a concern about a host's profile photo, or anything else. For a specific question, merge proposal, or host response, use the Report action on that item instead (long-press on the app, or the flag icon on the web) — it's faster and ties your report directly to that content.

Child safety concerns are reviewed by a person, not just logged. Selecting that category below sends an immediate alert to our designated child-safety contact in addition to being recorded.

Contact

General questions, complaints, or law-enforcement inquiries: admin@peanutforum.com

Designated child-safety point of contact: childsafety@peanutforum.com