EDITORIAL

Ethics Policy

Last updated October 11, 2026

RuntimeWire's editorial work is independent of its commercial and investment relationships. This page collects the standards we hold ourselves to.

Independence

Advertisers, sponsors, investors in RuntimeWire, sources, and subjects of our reporting do not get prior review of stories, and they do not influence what we choose to cover. Editorial decisions are made by the newsroom.

Conflicts of interest

Members of the RuntimeWire team are active operators and angel investors. When we cover a company in which anyone on our team has a financial interest — an investment, an advisory role, a paid relationship, a family tie, or another material connection — we disclose it in the article. When the conflict is significant enough that disclosure isn't sufficient, the story is assigned to someone else or not covered.

We keep a conflicts register: holdings, portfolio companies, sponsors, and close personal or business ties. Every story is checked against it before reporting starts, and each entry is handled one of three ways:

• Disclose. The article carries a disclosure line, added as fixed text so no edit can drop it.
• Second reviewer. The article carries the disclosure, and the person with the conflict cannot approve it; an editor without the conflict must.
• Recuse. The person with the conflict steps away. The story pauses until an editor decides who handles it, or whether it runs at all.

The approval record for each story notes which conflicts applied.

What we will not publish

• Fabricated sources or quotes.
• Undisclosed paid placements, gifted coverage, or pay-for-play.
• Misleading synthetic media presented as real footage.
• Unverified rumor presented as fact.
• Personally identifying information about private individuals when there is no clear public-interest justification.

Sources and anonymity

We default to on-the-record sourcing. Anonymous sources are used only when the information is significant, the source has a credible reason to require anonymity, and we can corroborate the claim. We protect source identities when we have committed to do so.

Fairness and right of reply

Before we publish a finding that is critical of a person or company, we put it to them with specific questions and a reasonable deadline: at least 24 hours, ending at 5 p.m. on a business day in their time zone. The story reports their response, or says they did not respond by the deadline. If we cannot reach them, an editor decides how they get a chance to respond before the story runs.

Accuracy checks before publication

• A story can fail. When the evidence is insufficient, the premise turns out to be wrong, or the finding is too weak for the headline, the reporter recommends holding or killing it, and an editor decides.
• Reporting looks for evidence against the story and for other explanations, and checks test conditions, sample sizes, denominators, and how far a finding generalizes.
• Quotes from calls are checked against the recording, not only the machine transcript. Numbers, names, and technical terms that the story depends on are re-heard from the audio, and a quote is never reconstructed from memory or a summary.
• Before approval, an editor sees the final draft with any problems the checks found, the questions still open, and the status of every claim. The approval is recorded against that exact version.

Source material and recordings

• Who has access. RuntimeWire editors, and the AI reporting system working on that story. Material a source gives off the record, or on condition of anonymity, is kept separate and is not given to the system that writes drafts. Background material is used only without naming or identifying the source.
• Call recordings. Calls are recorded only with the source's consent, given on the call. Recordings are deleted after 90 days, including from our telephony provider.
• Transcripts and emails. Kept as the evidence record behind published reporting, so quotes and claims can be checked and corrected later.
• Outside processors. Calls and recordings are handled by Twilio. Live call audio and transcription are processed by OpenAI. Emails are delivered by SendGrid. Drafting and checking run on AI model providers selected by the newsroom. These providers process the material to deliver their service; we do not sell source material.

Plagiarism and attribution

Other people's reporting gets credit and a link. We do not pass off others' work as our own. When a story originates with another outlet, we say so.

Editorial vs. commercial

Sponsored content, affiliate links, and paid partnerships are always labeled and visually distinct from editorial. We do not run native advertising designed to deceive.

Reporting an ethics concern

Email [email protected] with the subject "Ethics concern" and a short description. We take these seriously.