Fact checking software for journalists: check the story, then keep it right
Most fact checking software for journalists helps you test claims before a story runs. Veritas covers the other half of the job: it keeps re-checking the facts in stories you have already published and tells you which ones have gone wrong.
Key takeaways
- Newsroom tools such as Fact Check Explorer, Full Fact AI and ClaimBuster check claims before or while you report.
- After publication, office-holders change, statistics are revised and future tense expires. Nothing flags it.
- Veritas keeps re-checking published stories and sends editors sourced, confident findings with a suggested rewrite.
Reporters check facts under deadline, and the good tools for that are well known. But a story keeps making claims long after it is filed. The minister it names leaves office. The unemployment figure it quotes is revised. "The bridge will open next spring" becomes a sentence about the past.
This guide covers both jobs: the tools journalists use to check claims before and during reporting, and how to keep an archive of published work accurate. Tool descriptions come from each tool's own site, checked in October 2026.
Two different jobs: checking before you publish, and keeping it right after
You are testing claims made by others: a politician's statistic, a viral image, a quote in a press release. Speed matters, and the tools are built for searching, matching and tracing.
The question changes. Your story was right on the day it ran. Is it still right today? This job is easy to put off, because nothing flags it until a reader or a source complains.
Most roundups of fact-checking tools cover only the first job. A newsroom with a large archive needs both.
The tools journalists use to check claims today
These are the tools that appear again and again in newsroom guides. Each description is based on the tool's own site:
Searches fact-checks already published by fact-checking organisations, by keyword or by uploaded image, and shows who checked the claim and their rating. A companion API returns the same results for your own systems.
Monitors text, video, audio and social media for important statements, alerts you when a known false claim is repeated, and transcribes audio and video in real time. Full Fact says more than 45 organisations in 30 countries work with it, in English, French and Arabic. Pricing is by demo request.
A claim-spotting model from the University of Texas at Arlington that scores which sentences are worth checking. Its research paper describes a public API.
Runs tiplines on WhatsApp, Messenger, Telegram, LINE and Viber so audiences can send in claims. Meedan reports 58 tiplines and 143,631 unique users in 2025.
An open-source browser tool from Media Lab Bayern that pulls out a text's main points, finds documents and compares the two.
Google Lens shows other websites that carry the same or a similar image, a first step for tracing where a picture came from.
All of these are built for checking incoming claims and viral media. None of them describes re-checking your own back catalogue on a schedule.
What goes stale in a news archive
A correct story becomes an incorrect one in predictable ways:
"the health secretary", "the club's head coach", "the company's chief executive". Titles outlive the people who hold them.
Official figures are revised, and each new release makes last year's "latest figures" out of date.
"will open", "is expected to", "next year". The event happened, was cancelled or slipped.
"the tallest", "the biggest employer in the region", "a record high".
A quote later disputed, or a statement attributed to someone whose role has since changed.
These errors weren't mistakes on the day of publication, so no pre-publication check could have caught them. They sit in the archive, still ranking in search and still being read.
Why it matters: accuracy codes and audience trust
Press codes treat accuracy as an ongoing duty. Clause 1 of the Editors' Code of Practice (in force from 1 January 2025), which IPSO enforces, says a significant inaccuracy, misleading statement or distortion must be corrected promptly and with due prominence. The BBC's editorial guidelines say serious factual errors should normally be acknowledged and corrected quickly, and that in news, achieving due accuracy is more important than speed.
Readers notice. The Reuters Institute Digital News Report 2025 puts trust in news at 40% across its markets, unchanged for the third year. Asked what would raise trust, audiences named focusing on facts and checking stories before publishing among the top themes. When people want to check something that might be false, 38% say they would turn first to a news outlet they trust, ahead of fact-checkers at 25%.
That makes your archive a reference people rely on. An outdated sentence there costs trust in the same way a fresh error does.
See which facts on your site have gone out of date.
Request early accessHow Veritas fits a newsroom
Veritas is an autonomous fact-maintenance agent built for editors, standards desks and publishers with large archives. It runs in the cloud on a schedule.
New and updated stories come in from your sitemaps, Google News sitemap and RSS or Atom feeds.
Statistics, dates, who holds a role, quotes attributed to named people, events, comparisons and superlatives. Opinion and prediction are left alone.
Preferring primary sources such as statistics offices and official records. Your desk can list the source domains it trusts.
Supported, contradicted, outdated, misleading or unverifiable. "Outdated" means the claim was true when published and isn't any more.
Each with the evidence, citations and a suggested rewrite an editor can apply in one click. High-severity issues can go to Slack or a webhook.
Time-sensitive claims are re-checked on a schedule (fast-moving ones after 7 days) and unchanged claims are never re-checked twice. When an editor fixes the sentence on the live page, Veritas closes the issue itself.
Veritas also has a pre-publish check that fact-checks a draft before it goes live, and every editor decision teaches it which alarms were false.
A routine for a standards desk
Run new drafts through a pre-publish check alongside your usual reporting checks.
Work the queue of flagged claims on published stories, most-read stories first. Fix the sentence, cite the newer source and add a correction note where your policy calls for one.
Mark the finding "not an error" with a reason. Veritas uses those notes to stop raising the same false alarm.
Look at the pages whose health score (the share of checked claims that are not problems) is falling, and decide which to update properly.
Veritas exports every claim, piece of evidence, verdict and editor label as a versioned dataset in your own storage, which gives you an audit trail of what was checked and when.
For more on catching stale claims across a whole site, see our guide to finding outdated content.
Tool lists go out of date too
Choosing software from a roundup has its own accuracy problem. Several guides still ranking in October 2026 describe a fact-checking landscape that has moved on:
- Meta ended its third-party fact-checking programme in the United States in January 2025, according to the Duke Reporters' Lab, yet some guides still present it as current.
- The Reporters' Lab's 2026 census reports that Logically Facts stopped publishing in 2025, while at least one 2026 guide still recommends Logically.
- One guide still says Full Fact's software isn't ready for the public. Full Fact's own site now describes it as in use across 30 countries.
The wider field is shrinking: the Duke Reporters' Lab counted 437 active fact-checking projects as of 1 June 2026, down from 441 at the end of 2025, with more than 30 projects closing in 2025 and 10 starting. Check a tool's own site before you rely on a list.
Honest limitsWhen Veritas isn't the right fit
Full Fact AI's monitoring or a Meedan Check tipline is built for that.
Reverse image search and dedicated media-forensics tools do that job. Veritas checks text claims.
Google Fact Check Explorer is built for exactly that.
Veritas reads the HTML a page serves.
See how Veritas works or the use cases it covers.
How Veritas works
- Finds your pagesReads your sitemaps, Google News sitemaps and RSS or Atom feeds on a schedule.
- Pulls out the claimsStatistics, dates, who holds a role, quotes, events. Opinion is skipped.
- Checks the evidenceSearches outside your site and prefers primary sources.
- Hands editors the fixConfident findings only, with citations and a one-click rewrite.
Frequently asked questions
What fact checking software do journalists use?
Common choices are Google Fact Check Explorer for finding existing fact-checks, Full Fact AI for monitoring claims across media, ClaimBuster for spotting check-worthy sentences, Meedan Check for audience tiplines and reverse image search for pictures. Veritas is for a different job: keeping your own published stories accurate over time.
Can AI fact-check a news story?
AI can find checkable claims, search for evidence and suggest a verdict, but an editor should make the call. Veritas shows only findings that clear a confidence threshold, always with the evidence and citations, and an editor decides whether to apply the suggested fix.
Is there free fact checking software for journalists?
DoppelCheck is open source, and Google Fact Check Explorer and Google Lens are open to anyone with a browser. Check each tool's own site for current terms.
How do newsrooms keep old articles accurate?
Without a routine, problems tend to surface only when a reader or source complains. A better approach is to re-check time-sensitive claims (office-holders, statistics, dated phrasing) on a schedule, fix the most-read stories first and log each correction. Veritas automates the re-checking and the queue.
Does an outdated fact in an old story need a correction?
It depends on your policy and on whether the story was wrong when published. The Editors' Code requires significant inaccuracies to be corrected promptly and with due prominence. A sensible approach is an update note when a fact has changed since publication, and a correction when the story was wrong at the time.
Can Veritas check a draft before it is published?
Yes. Veritas has a pre-publish check that fact-checks a draft before it goes live, using the same claim extraction and evidence search it uses on published pages.
Keep every page true
Veritas re-checks the facts across your published pages and hands editors a sourced fix when one goes stale.
Request early accessRelated guides
- A fact checking tool for editors who answer for everything that's publishedMost tools sold as a fact checking tool for editors check one draft at a time. Editors are responsible for much more: copy from many contributors, headlines, corrections and every story already in the archive. Veritas keeps checking that published work and puts the problems in one queue.
12 sources, checked 29 October 2026
- Editors' Code of Practice Committee: The Code, clause 1 Accuracy (in force 1 Jan 2025)
- BBC Editorial Guidelines: Accuracy (seen 2026-10-29)
- Reuters Institute: Digital News Report 2025, executive summary
- Duke Reporters' Lab: 2026 fact-checking census (12 Jun 2026)
- Duke Reporters' Lab: fact-checking census posts (2025 census)
- Google News Initiative: Google Fact Check Tools training
- Google: Fact Check Tools API
- Full Fact AI (seen 2026-10-29)
- ClaimBuster paper, arXiv 2002.07725 (2020)
- Meedan Check (seen 2026-10-29)
- Media Lab Bayern: DoppelCheck
- Google Search Help: Search with an image