Privacy policy · Terms of use · Community guidelines · How moderation works · Research organisations
Sp-Otter · Singapore · Updated 25 October 2026
Before anything is published, every photo and caption is checked by AI (open models run by Cloudflare, or Anthropic’s Claude, plus Google Cloud Vision when it’s switched on). The checks look for unsafe content, personal information and whether an otter is in the picture.
Unsafe content is rejected straight away. Clear otter photos are approved and appear on the map after 15 minutes. If the checks aren’t sure it’s an otter, the post is never rejected automatically. It goes to trusted spotters instead. When the checks run on open models (Cloudflare), photos that aren’t clearly safe aren’t rejected automatically either: a moderator looks at them first, and only a person’s decision can give a strike.
Three “otter” votes publish a post, three “not an otter” votes reject it, and two “unsafe” flags remove it. After 24 hours, two otter votes with no objections are enough. A moderator’s vote decides on its own, and photos that may be unsafe are shown only to moderators.
New spotters get the strictest checks and up to 3 posts a day. After 2 approved sightings you become a Spotter. After 8 approved sightings and 14 days you become a Trusted spotter and can help verify posts.
Anyone can report a post or comment, and it disappears for them at once. Enough reports hide it for everyone automatically (trusted spotters’ reports count double). A report from an established account (at least 7 days old, no strikes, up to 10 reports a day) that hasn’t led to action within 12 hours hides the post or comment for everyone until trusted spotters or a moderator review it; nothing is deleted that way. Breaking the rules earns a strike. Three strikes within 90 days pauses the account until they expire.
If the automatic checks are briefly unavailable, nothing unchecked is published. Your sighting waits privately — only you can see it, marked “pending review” — and is checked automatically as soon as possible. If it still can’t be checked after 7 days, it’s deleted and you’re told.
Comments get the same automatic safety check before they appear.
Each automatic or community decision is recorded with its scores so mistakes can be found and the checks improved.
Questions: [email protected]