What we are, and why we’re careful.
Casepin is a world map of documented crimes. Every pin opens a neutral, sourced case file (what happened, a timeline, the people involved with their real legal status) with the best existing coverage attached.
It is built victim-first, for adult readers, as a reference rather than a feed.
- 3,153
- 92
An ethical discovery layer.
True crime is enormous and scattered across videos, threads and articles of wildly uneven quality. Casepin is the layer that organizes it.
- Explore by place.Every case sits on the map where it happened. You browse geography (a city, a coastline, a street), not an algorithmic feed built to keep you scrolling.
- Read the verified record.Each pin opens a neutral case file: what happened, a timeline, the people involved with their real legal status, and the sources the file was written from.
- Find responsible coverage.The best documentaries, videos and podcasts on a case are linked and embedded in context, credited to their makers, and never re-hosted.
The genre usually gets this wrong.
Much of true crime treats real deaths as entertainment. Cases get sensationalized, victims get flattened into plot, and people who were never charged get named by strangers online. Casepin exists to be the deliberate opposite: a calm, accurate reference that keeps the victim at the centre and sends attention to the people who cover these cases responsibly.
In this genre, restraint is the whole point. What we choose not to build matters as much as what we do.
Documented, sourced, and gated.
It has to be on the record.
A case earns a place only if it is documented in public reporting or records. The catalog focuses on the United States, the United Kingdom, Canada and Australia, plus globally notorious historical cases. Ongoing trials and very recent cases are held to a higher bar or deferred.
Written from sources we already hold.
A first draft is written only from source text we have retrieved: published journalism, court and government records, and encyclopedic sources. Videos and podcasts are coverage we point you to, not what we cite facts from.
Two citations, at least.
No case file publishes with fewer than two independent citations, and every one is listed on the case page. Automated checks require valid citation URLs from independent hosts, valid dates, permitted language, and supported legal-status labels for named people.
Gated before it appears.
Files are written as drafts, then move through a review queue before anything goes live. Only files marked published are ever visible. If a file is later flagged or pulled, it leaves public view straight away.
Corrected in the open.
Each file shows the date it was last verified against its sources. When facts change, the file is rewritten from updated sources and marked with the date, and every correction goes into the public log.
How we use AI, exactly.
We use AI, and we are precise about how. On the site, language models write a first draft of a case file from source material we have already retrieved, and they help match existing coverage to the right case. On Casepin, a synthetic voice can read the already-published dossier text aloud; that reading is not a new source or a separately generated factual account. Synthetic voices also narrate our short- and long-form videos, atmospheric illustrative stills may pad a case’s real archival records, and software helps assemble and edit those videos.
Those drafts use only the source text we pass in. They do not use the model’s own memory or the open web. We instruct the model to add no fact that is absent from those sources. If the pipeline cannot tie a named person to both a citation and a legal status, it removes that person from the draft.
The model never decides what goes live. Publication is under owner control. Automated checks require valid citation URLs from independent hosts, valid dates, permitted language, and supported legal-status labels for named people. Holds and high-risk files go to the human review desk; owner-authorized batches must pass the same machine floor at publication time, and failures remain held. A human-run corrections process continues after a case is public.
Generated stills use atmospheric illustrations. They do not depict a real victim, perpetrator, or crime scene. We disclose this here and on our standards page instead of adding a visible AI label to each post. We set a platform’s machine-readable AI flag only when that platform requires it.
If we get something wrong.
A correction link sits on every case file, and anyone can use it. We aim to give a first response within 72 hours, we log every request, and when facts change we rewrite the file from updated sources and mark the date it changed.
For families
- Open the correction form from any case file, or from the link below. Tick “I’m a member of the victim’s family.”
- Family requests go to the front of the queue, and a person reads every one. We aim to give a first response within 72 hours.
- We fix errors quickly, and we will consider de-emphasizing even accurate content when a family asks. De-emphasized files leave every featured surface.
- Full removal is decided case by case, by a person, never automatically.
Who runs Casepin.
Editorial team
Casepin editorial desk
Casepin is an independent project. Details about who runs it are being finalized and will appear here: a name, a short note on the person behind it, the operating entity, and a way for press to make contact.
Reach a person.
- The submission desk is the best way to reach us: suggest a case, point us to better coverage, or flag something that needs fixing. Every message is read by a person before anything changes on the site.
- For press, a full kit (boilerplate, fast facts and quotable positioning) lives on the press page, with a press contact and an asset pack available on request.