Code of Standards.

We publish to inform the public. Every claim we publish is one we believe to be true at the time of publication and one we are prepared to defend in full. Where evidence is partial, we say so. Where it is contested, we show the contest. We use precise, factual language, take care not to overstate findings, and do not publish to entertain, to provoke, or to perform.

All Signal & Shadow outputs are grounded exclusively in observable, verifiable data. Four constraints apply without exception.

The mandatory substitutions are as follows.

Every visual, document, account, and dataset we publish has been verified using methods consistent with the Berkeley Protocol. We confirm location, time, and provenance by independent means before attribution. The central claim of an investigation is supported by at least two independent sources, except in cases where only one relevant source exists, in which case we say so. We rely on primary sources wherever suitable primary sources are available and explain any reliance on secondary sources. We preserve the original evidence, the verification trail, and the analytical reasoning behind each finding. We do not republish material we cannot verify.

Every source used in a Signal & Shadow output is assigned an Admiralty 6x6 grade before the output is drafted. The grade is expressed as a letter for source reliability and a number for information credibility.

Counter-narrative discrepancies, defined as instances where high-reliability sources of grade A or B provide conflicting metadata or timestamps, are flagged explicitly in the Admiralty Grade Triage section of every dossier.

All factual claims are tagged with one of three provenance categories.

All media artefacts examined under the Signal & Shadow framework are subject to the Media Integrity Protocol. The protocol assesses file metadata status, recording METADATA STRIPPED BY HOST PLATFORM where applicable; artefact detection through ELA variance and FFT inconsistencies; codec fingerprinting; chronological markers; C2PA Content Credentials status reported as VALID, STRIPPED, or UNSIGNED; and Hard Bindings (SHA-256) and Soft Bindings (perceptual hash) for file matching and variant tracking.

Every Signal & Shadow forensic output includes a technical appendix sufficient for an independent analyst to reproduce the methodology. The appendix contains tool logs and query strings used during analysis, raw SHA-256 hash values for all media artefacts examined, SunCalc or equivalent chronolocation calculations where applied, satellite imagery cross-references with source, date, and resolution, and EMS-98 structural compromise assessments where relevant.

No investigation is published without review by at least one person other than the author. The reviewer tests the evidence, the reasoning, the framing, and the language against this code. The reviewer is named in our internal record for each published work and is accountable for the review.

We identify our sources unless doing so would expose them to harm, in which case we explain why anonymity has been granted. Anonymous sources are used only where the information they provide is corroborated by named sources or by material evidence. We do not pay sources for information. We disclose when material has been provided by an interested party, when access has been mediated, and when claims rest on a single source. When citing experts, we establish their relevant credentials and disclose any reasonable conflict of interest they hold on the subject in question.

We offer a right of reply, where appropriate and feasible, to any named person or organisation that is the subject of significant criticism, allegation, or attribution in our work. We allow reasonable time for response, we represent the response fairly in the published work, and we record the offer and any response received in our internal file for the investigation.

Open source material is not consequence-free. We assess the risk our work poses to identifiable individuals before publication, with particular care for civilians, witnesses, minors, survivors of violence, and people in jurisdictions where exposure carries legal or physical danger. We redact, blur, or withhold where publication would cause foreseeable harm that outweighs the public interest. We do not publish the faces, names, or locations of children except where the public interest is clear, exposure is unavoidable to the story, and the editor has signed off in writing. We take measures to avoid exposing the original poster of user-generated material to undue scrutiny where they are not themselves the subject of legitimate investigation.

We collect the minimum data necessary to verify and publish. We store evidence securely, with access limited to those working on the investigation. We maintain chain of custody records for material that may have evidentiary value in legal or accountability proceedings. We anonymise personal data that could enable harassment of individuals, including telephone numbers, identification documents, financial details, and private contact details. We do not retain personal data beyond the period required for the work and any reasonable archival purpose.

No Signal & Shadow workflow compromises source anonymity or exposes personally identifiable information without a clear, documented public interest justification.

Operational security is maintained during all technical discovery, including the use of dedicated virtual machines, passive lookup tools, and appropriate network hygiene. Source anonymity is protected at all stages. Local language models or privacy-mode processing are used when handling material that contains personally identifiable information or sensitive source data.

We use AI tools as aids to research, analysis, drafting, and translation. We do not publish AI-generated imagery, audio, or video as evidence. We do not present AI output as a source. Any AI-assisted geolocation, transcription, classification, or translation is verified by a human analyst before it informs a published claim. We disclose material AI use when relevant to the reader's assessment of the work.

Signal & Shadow is editorially independent. No funder, client, advertiser, partner, or platform has any right of review, approval, or veto over our investigations. We do not endorse political parties or candidates for public office. We do not enter agreements or partnerships with political parties, campaigns, or government bodies that would compromise our independence. We do not focus our investigations unduly on any one party or side of the political spectrum.

No person working on a Signal & Shadow investigation may hold a salaried or prominent position in a political party, campaign, government, or state-controlled body. Staff and contributors do not accept gifts, favours, or services with advantageous conditions that go beyond ordinary courtesy. We disclose personal, financial, professional, and political relationships that could reasonably be seen to affect judgement on a given subject. Where a conflict cannot be managed by disclosure, the affected person recuses themselves from the work.

We publish a current statement of funding at signalandshadow.io/funding. The statement names every source contributing more than 5% of annual income or €5,000, whichever is lower, for the most recent two financial years. It separately identifies all income from online platforms, public bodies (national or international), and any other source that may give rise to a perceived conflict of interest. Where an investigation has been supported by a specific funder, that funder is disclosed in the published work. We acknowledge in-kind support, including free or discounted access to tools, software, datasets, or training, where it makes a material difference to our work.

Contributors to Signal & Shadow accept this code as a condition of publication. Contributors retain authorship of their work and are credited by name unless safety considerations require otherwise. We pay contributors fairly and on time. We do not exploit early career, freelance, or locally based investigators through unpaid labour, uncredited work, or unmanaged risk transfer. Contributors are briefed on this code, on our verification standards, and on the safety and conflict-of-interest provisions before commission.

All Signal & Shadow outputs are framed as independent forensic audits or verified investigative dossiers. No Signal & Shadow output is framed as a submission to law enforcement or judicial authorities. Signal & Shadow operates as a private technical authority with evidentiary standards designed to withstand independent scrutiny.

All Signal & Shadow outputs are produced in British English and adhere to AP Style. Sentence case is used for all headlines and subheadlines. Terminal punctuation is not used in headlines. Plain language is preferred. Technical jargon is defined on first use.

We correct every known error of fact promptly, openly, and in full. Each correction is appended to the original work, dated, and explained. Where the original cannot be edited, we issue the correction in the same channel as the original. Where an error is substantial, meaning it changes the conclusion of the investigation or departs from this code, we set out the accurate finding, an account of how the error occurred, and the steps taken to prevent recurrence. We do not silently amend published material. Typographical and stylistic fixes do not require a correction note. A central register of corrections is maintained at signalandshadow.io/corrections, updated as corrections are issued.

Any person may raise a complaint about our work by writing to [email protected]. We acknowledge receipt within five working days and respond substantively within twenty working days. Where a complaint is upheld, we publish the outcome alongside the affected work and on the corrections register. Where we disagree, we explain our reasoning. Complainants who remain dissatisfied may escalate to the independent reviewer named at signalandshadow.io/code-of-standards.

We work within applicable law, including data protection, copyright, and national security frameworks in the jurisdictions where we operate and publish. We do not commission, encourage, or accept material obtained through unlawful access to computer systems, intercepted communications, or coercion.

This code is reviewed at least annually and revised as practice develops. Material changes are versioned and dated. Current and superseded versions are archived at signalandshadow.io/code-of-standards. Readers may propose amendments by writing to [email protected].