Acceptable use policy

A tool that maps claims, connects entities and reconstructs chronologies across a document set is useful to a journalist checking a filing and equally useful to someone assembling a case against a neighbour. The difference is not technical. It is a matter of what the material is, who it concerns, and what is done with the result.

This policy is part of the terms of service. Breaching it is a breach of those terms.

You must have the right to hold and process everything you put into a case. Before uploading, satisfy yourself that you do.

Do not upload:

• Material obtained through unauthorised access to a system, account or premises.
• Material you are barred from disclosing to a third-party processor by a protective order, a confidentiality undertaking, a sealing order or the terms of a records release.
• Special-category personal data — health records, biometric data, information about sexual life, religious or political affiliation — unless you have a lawful basis and a genuine investigative need.
• Material whose possession is itself unlawful in your jurisdiction, including sexual content involving minors.
• Malware, or files crafted to attack the processing pipeline or another workspace.

Where material is subject to a legal obligation that travels with it, that obligation remains yours. CaseSignal processes what you give it on your instructions; it cannot assess whether you were entitled to give it.

Investigations name people. The distinction that matters is between examining conduct that a person carried out in a public or official capacity, and building a picture of a private individual’s life.

You must not use CaseSignal to:

• Harass, threaten, intimidate or stalk anyone, or assist anyone else in doing so.
• Compile a dossier on a private individual — a neighbour, a former partner, a colleague, a critic — who is not the legitimate subject of an investigation into conduct in a public, professional or official role.
• Publish or assemble for publication a person’s home address, personal contact details, movements, family members or other material whose disclosure exposes them to harm. This applies to public figures too: scrutiny of a role is not licence to expose a household.
• Attempt to de-anonymise anyone — to identify a confidential source, a whistleblower, a protected witness, a pseudonymous account holder, or a person a redaction was applied to protect — including by correlating records to defeat the redaction.
• Build profiles or make inferences about individuals on the basis of race, ethnicity, national origin, religion, health, disability, sexuality, gender identity, immigration status or political affiliation.
• Support surveillance of a person that would be unlawful if you carried it out directly.

CaseSignal produces descriptions of what documents state. Those descriptions carry no authority beyond the documents behind them, and they must not be dressed up as though they do.

You must not:

• Use an output to allege, state or imply that a named person or organization committed a crime as an established fact. A claim status describes cited excerpts; a discrepancy describes a difference between documents. Neither is a finding of guilt, fraud, misconduct or intent.
• Present an output as a certified, official, expert or independently verified determination, or as the work of a court, regulator, auditor or law-enforcement body.
• Publish a finding without checking the citations behind it against the underlying records.
• Strip citations from an output in order to present a claim as more settled than the record supports, or edit an excerpt and continue to present it as verbatim.
• Publish an evidence room containing material you are not permitted to disclose, or that identifies a person the records were meant to protect.

The URL importer exists so that a public page cited in an investigation can become a citable source alongside the documents. It is a tool for adding specific pages you have identified — not a collection mechanism.

You must not:

• Crawl a site, enumerate its URLs, or import pages at machine speed or in bulk.
• Script or automate the importer to assemble a dataset from a third-party site.
• Use it to reach material behind a login, a paywall or an access control you are not entitled to pass.
• Use it against a site whose terms or robots directives prohibit automated retrieval, or in a way that burdens the site being fetched.
• Point it at internal, private or metadata endpoints, or attempt to use it to reach a network you could not reach directly.

Requests that resolve to private, loopback or link-local addresses are refused, redirects are re-validated on every hop, and import volume is rate limited. Those are safeguards, not a substitute for this rule.

Do not interfere with the service or with other people’s use of it. Specifically, do not:

• Circumvent or attempt to circumvent rate limits, plan limits, authentication, authorization checks or any other security control, including by creating multiple accounts to defeat a limit.
• Access, or try to access, a workspace, case, file or share link that is not yours.
• Scan, probe, fuzz or load-test the service except within the terms of responsible disclosure, and never against another user’s data.
• Reverse engineer the service to build a competing product, or resell or sublicense it without a written agreement.
• Use the service in a way that degrades it for others, or that is designed to run up cost rather than to do investigative work.

This list is a summary of the sections above and does not replace them. It is not exhaustive; conduct that is plainly against the spirit of this policy is against the policy.

• Uploading material you have no right to hold, share or process.
• Harassing, threatening, intimidating or stalking any person.
• Compiling dossiers on private individuals who are not the subject of a legitimate investigation.
• Publishing home addresses, personal phone numbers, family details or other identifying material about a private individual.
• Attempting to de-anonymise a source, whistleblower, witness or any individual protected by redaction or pseudonymisation.
• Presenting an output as an established finding that a named person or organization committed a crime.
• Using CaseSignal to build profiles based on race, ethnicity, religion, health, sexuality, immigration status or political affiliation.
• Automated bulk collection through the URL importer, including crawling a site or importing at machine speed.
• Circumventing rate limits, access controls, authorization checks or any other security control.
• Probing, scanning or load-testing the service outside responsible disclosure.
• Uploading malware, or content designed to disrupt processing or another user’s workspace.
• Reselling, sublicensing or white-labelling the service without a written agreement.

We do not monitor the content of cases, and we do not read your source material as a matter of routine. When we receive a credible report, we investigate it to the extent necessary and act proportionately.

Depending on what we find, we may ask for an explanation, require specific material to be removed, disable a published evidence room, suspend an account, or terminate access. Where harm is ongoing or serious we may act first and explain afterwards. Where we are legally required to act, we will.

If your access is suspended over something you believe was a misunderstanding, tell us and we will look again. Where practical and lawful we will allow you to export your cases before access ends.

A public contact address has not been configured for this deployment, so reports of misuse should be sent to the operator running it. Security vulnerabilities go through responsible disclosure instead.

If you are named in material published through a CaseSignal evidence room and believe it breaches this policy, tell us. You do not need an account to make that report, and we will not require you to identify yourself beyond what is needed to look into it.

Related documents: the terms of service and the privacy policy.