Skip to content

AI for journalists who cannot upload leaked documents

Short answer

A leaked document in a cloud tool is a document on somebody else's server, subject to their retention policy, their jurisdiction and their response to a subpoena.

The rule usually cited
source protection obligations
The document in question
leaked documents

What the rule actually turns on

This is the case where the threat model is not abstract. Material held by a third party can be compelled from that third party without you ever being told. A local model means there is no third-party copy to compel — the only copy is on your machine, where your own operational security governs. That is a genuine difference in exposure rather than a policy promise.

What journalists use it for

  • Reading a large document dump for names, dates and amounts
  • Summarising a long report before a deadline
  • Working with material on an air-gapped machine
  • Checking a transcript against claims made in a statement

The caveat for this work specifically

Device security is now the whole of your security. A local model protects the document from a vendor, not from a seized laptop.

This is not compliance advice

The rules named here are real and our reading of them is not authoritative. What this page describes is a technical fact — that nothing is transmitted — and what that fact does to the analysis. Whether your regulator, your bar, your IRB or your security officer is satisfied is their decision, and the claim is checkable before you take it to them.

Other ways people ask this

  • ai for journalists
  • can journalists use chatgpt
  • confidential ai tool for journalists
  • I'm a journalist. Can I use AI on leaked documents without breaching source protection obligations?
  • What AI tools can journalists use on leaked documents when cloud upload is not allowed?

Related