Direct answer
What is Full Fact AI?
Full Fact AI is a specialized software suite created by the UK fact-checking charity to monitor media streams, isolate verifiable claims, and match repeated misinformation.
Full Fact AI is a suite of machine-learning tools developed by Full Fact, the UK's independent fact-checking charity, to scale up the detection and verification of misinformation. Operating since 2016, the system handles media monitoring across online news sites, speech from live TV, podcasts, and social media channels. Data is ingested into a user interface where text is segmented into individual sentences for processing. The system uses a claim-type classifier powered by a fine-tuned BERT model, trained on Full Fact's annotated dataset. This classifier isolates checkable statements—such as claims about quantities or cause-and-effect—while filtering out unverifiable statements like predictions. Claims are then categorized by topics, such as health or the economy, helping editorial teams triage large volumes of information. To identify repeats of previously addressed falsehoods, Full Fact AI employs a matching pipeline. This combines sentence vectorization, entity analysis, and generative AI models to detect matches even when claims use different wording. The tools do not automate verdicts; rather, they surface candidates for human journalists to investigate, whose findings can be published using ClaimReview markup for integration into search engines like Google.
Unlike generic transcription or LLM summarisation products, Full Fact AI is custom-engineered by frontline fact-checkers to isolate checkable factual assertions and trace repeat appearances of previously debunked claims across multimodal feeds.
Product capabilities
Key features
Multimodal Data Ingestion
Monitors and extracts text from online news sites, live television speech, audio podcasts, and social media channels into a unified interface.
Sentence-Level Claim Labelling
Applies a fine-tuned BERT model to isolate checkable factual assertions—such as quantities and causality—while filtering out unverifiable predictions.
Topic Categorisation
Sorts identified claims into topic areas like the economy and health to streamline review for specialized beat journalists.
Repeat Claim Matching
Uses sentence vectorization, entity analysis, and generative AI to surface repetitions of previously fact-checked claims despite phrasing variations.
ClaimReview Schema Integration
Assists publishers in marking up completed human fact checks using schema.org ClaimReview structured data for search engine visibility.
Practical fit
Who should use Full Fact AI?
Broadcast and Podcast Monitoring
Extracting and transcribing long-form audio to quickly flag suspicious or misleading health claims without manual listening.
Tracking Political Repeat Claims
Alerting journalists when public figures repeat specific misleading statistics that have already been scrutinized and debunked.
Election Triage
Managing high daily sentence volumes during electoral cycles to keep pace with rapid political discourse and media coverage.
Editorial assessment
Pros and limitations
Where it is strong
- Built and battle-tested by an accredited, professional fact-checking organisation.
- Filters out non-checkable language such as future predictions to reduce cognitive fatigue.
- Combines entity extraction and generative models to match claims with varying wording.
Where to be careful
- Does not automate the fact-checking verdict itself; requires human editorial investigation.
- Access is restricted to paid institutional licensing and is not openly available for self-service sign-up.
- Documented daily operational support is focused on selected languages rather than universal global coverage.
Commercial context
Full Fact AI pricing
As of September 2026, Full Fact AI is provided to organisations via a paid licence upon enquiry. Public pricing tiers are not published on the site, so potential users should contact Full Fact directly for terms.
Pricing, limits, taxes, model access, and regional availability can change. Verify the purchase-critical details on the official pricing page linked under Sources.
Transparent ranking
Why Full Fact AI scores 70.9
Each factor is scored on a 100-point scale, then combined using the public ToolsRank weights. Engagement and momentum stay at a neutral baseline until measured signals exist, so no tool can gain or lose position from numbers nobody recorded.
Compatibility
Languages, platforms, and integrations
Languages
- English
- Arabic
Integrations & surfaces
- ClaimReview (schema.org)
- Google Fact Check Explorer
Community
Reviews and questions
No approved member reviews yet. Editorial factors above are the only rating on this page.
Reviews and questions come from Google-signed members and are checked by an editor before they appear.
Frequently asked
Full Fact AI FAQ
Does Full Fact AI automatically determine whether a claim is true or false?+
No. Full Fact emphasizes that the AI does not replace human experts. The technology collects data, isolates checkable sentences, and flags repeats of previously checked claims, but the actual investigation and adjudication are conducted offline by human fact-checkers.
What kind of media can Full Fact AI monitor?+
According to the official documentation, users can configure the system via a simple UI to monitor online news websites, live television speech, podcasts, and social media pages.
How does the claim classifier work?+
Full Fact AI uses Google Research's BERT model, fine-tuned on Full Fact's proprietary annotated dataset. It categorizes checkable claim types—such as figures and cause-and-effect statements—while filtering out unverifiable predictions.
How can an organisation get access to Full Fact AI?+
The tools are available to external organisations under a paid licence. Interested newsrooms, civic groups, and platforms must reach out to Full Fact directly to arrange testing or licensing.
