Content policy

What TruthSeekers shows

Every question gets AI-written lenses grounded in retrieved sources. Each one is always marked with a spark icon (labelled “Written by AI from the cited sources” for screen readers and on hover) and always cites what it draws on. Community members can add their own viewpoints and evidence underneath; those carry a separate people icon (“Submitted by a community member”) and start hidden until a moderator approves them.

Showing a viewpoint is not an endorsement of it. Presenting an alternative view alongside the mainstream one is the point of the product, not a claim that both are equally likely to be true — see the evidence-strength labelling below for how we handle topics where they clearly aren’t.

Evidence-strength labelling

Each topic is classified one of three ways, shown on the results page:

• Contested — genuine open disagreement (e.g. who was responsible for a historical event with disputed facts). Shown as two lenses, given equal space.
• Settled, with dissent — the evidence clearly points one way, but an organised fringe claim exists (e.g. the moon landing). We say so with a banner, and the alternative lens is reframed to address the dissenting claim directly rather than presented as an equal possibility.
• Settled atrocity — a documented genocide or comparable atrocity with an organised denial campaign, on a short, deliberately narrow list reviewed by a person, not decided by the AI alone. We lead with the documented record; denial claims appear only in a clearly labelled, evidence-answered section, never as an equal lens.

A wrong classification is a real failure mode we take seriously. If you think a topic is mislabelled, use the “Report a problem with this” link on the lens itself.

What’s not allowed in community submissions

Evidence and viewpoints are removed, and repeated violations can lose posting privileges, for:

• Hateful or harassing content directed at a person or group
• Spam, or content unrelated to the topic it’s posted under
• Deliberately misleading claims, broken or spoofed links, or a source misrepresented as primary when it isn’t
• On a settled-atrocity topic specifically: content promoting denial of the documented record

How reporting works

Community evidence and viewpoints can be flagged directly (broken link, misleading, not a primary source, hateful or harassing, spam, off-topic). Three open flags from three different people send an item back to a moderator queue automatically.

AI-written lenses aren’t flagged the same way — they have no author and nothing to send “back to pending.” Instead, each lens has its own “Report a problem with this” link (inaccurate, unfair or biased, hateful or harassing, unsafe or harmful, or something else). Reports go to a moderator queue for review; they’re a signal, not an automatic takedown.