Selected for the STM Innovator Fair 2026 | London, 8 December

Detect AI generated errors before publication.

AI can produce convincing scientific claims that misrepresent findings from real publications. These errors can be subtle, difficult to detect and scientifically significant.

RefCheck specializes in verifying scientific claims against their cited sources using direct textual evidence and repeated LLM assessments (currently available for life sciences and biomedical manuscripts).

One check, done in depth: does the source support the claim?

AI can cite a real paper and still get the science wrong.

64%

of researchers express concerns about AI inaccuracies and hallucinations.

Wiley, 2025
8.0%

estimated major quotation errors in medical literature.

Baethge & Jergas (2025), Research Integrity and Peer Review
Not all researchers use frontier AI models, and error rates vary across models. Even as advanced systems improve, subtle distortions of real scientific findings remain difficult to detect.

The harder errors are hidden in legitimate references.

A numerical change, an overstated causal relationship or an unsupported conclusion can alter the scientific meaning.

RefCheck. One check, done in depth.

RefCheck verifies scientific claims against their cited sources using direct textual evidence and repeated LLM assessments.

See exactly where the claim differs from the source.

RefCheck example finding
Real RefCheck finding: Van Dongen et al. (2003), manuscript claim of four nights compared with source passage reporting one or two nights of sleep deprivation
[1] Van Dongen et al. (2003) Inflated figure

WHAT THE MANUSCRIPT CLAIMS

The study found that restricting sleep to 6 hours per night for 14 days produces cumulative cognitive deficits equivalent to four full nights of total sleep deprivation [1].

WHAT THE SOURCE SAYS

Passage source: Abstract

“Chronic partial sleep restriction of five to six hours per night produces cumulative cognitive deficits comparable to one or two nights of total sleep loss”

Tap the image to view the complete finding at full size.

Click or tap to expand the full finding

What makes each finding actionable

Side-by-side evidence

The manuscript claim and original source passage.

Color-highlighted discrepancies

The exact information that does not match.

Scientific explanation

What is wrong and why it matters.

Editorial guidance

What needs review or clarification.

How do we achieve this?

RefCheck spots the hardest errors by combining independent source evidence extraction with a multi-layer LLM-as-judge architecture, repeated assessments and additional verification of difficult cases.

How RefCheck Verifies Claims
RefCheck also verifies reference existence and bibliographic metadata.
View a Sample Report

No false alarms. Errors detected.

0%

False positives on supported claims

100%

Of errors detected when the cited source was available

Development evaluation results; not yet measured on the held-out Gold TEST.

How RefCheck Verifies Claims

Evaluate RefCheck on your manuscripts.

Pilot evaluations are scoped and priced for each publisher.

Scope

We agree on journals, number of manuscripts, evaluation criteria and success metrics.

Run

Your team uploads and analyzes manuscripts directly in RefCheck, receiving an automatic evidence-backed report for each.

Review

We review the findings with your editors and deliver a consolidated pilot report with results and recommendations for next steps.

Discuss a publisher pilot

Build the next generation of scientific verification with us.

Our AI Truthfulness research investigates internal signals of scientific evidence fidelity in language models. We collaborate with publishers to develop specialized, efficient verification tools.

Read the research

High-volume screening

Develop lightweight classifiers to quickly flag claims that may not be supported, at lower cost.

17%Fewer LLM judge evaluations in development experiments.

Publisher-specific classifiers

Research and develop models tailored to your disciplines, scientific error types and annotated editorial data.

Agentic verification for complex cases

An orchestrator agent coordinates additional evidence retrieval, extraction and LLM assessments for cases that remain unresolved, reserving deeper analysis for the claims that need it.

Your full-text catalogue

Use authorized scientific content to support deeper evidence verification and the development of specialized methods.

Explore a development partnership

Bring deep scientific verification into your editorial workflow.

RefCheck can complement existing research integrity platforms with specialized claim-to-source verification. We welcome partnerships to explore integration with editorial systems, including ScholarOne Manuscripts, Editorial Manager and the STM Integrity Hub.

Discuss a technology partnership

Let's discuss scientific verification for your organization.

Eduardo Fuentes, PhD

Inventor of RefCheck
CEO of WriteWise

Discuss publisher pilots, collaborative research and technology partnerships directly with me.

Visit my website

We will use your contact details to respond to your RefCheck inquiry.

Frequently asked questions

RefCheck example finding
Real RefCheck finding: Van Dongen et al. (2003), manuscript claim of four nights compared with source passage reporting one or two nights of sleep deprivation