SourceVerify

• AI-powered citation verification that checks whether references actually exist and match their cited details
• Field-by-field verification of titles, authors, publication years, venues, DOIs, URLs, and other identifiers
• Detection of fabricated, hallucinated, incomplete, or inaccurate academic references
• Automatic repair of incorrect citation metadata, including authors, years, titles, venues, and identifiers
• Bulk verification of complete reference lists and bibliographies containing hundreds of citations
• PDF reference verification for extracting and checking citations from uploaded academic documents
• Evidence-based verification results showing matched fields, discrepancies, repairs, and supporting evidence
• SVRIS open verification standard providing deterministic and auditable citation classifications
• Reference formatting for APA, MLA, Chicago, and other academic citation styles
• Browser extensions, REST API, CLI, and MCP integration for citation verification across research workflows

What Is SourceVerify?

SourceVerify is an AI-powered reference verification platform designed to help researchers, students, reviewers, publishers, and research organizations check whether citations are genuine and accurately represented. The platform focuses specifically on reference integrity rather than general-purpose writing or plagiarism detection.

SourceVerify checks whether a cited work exists, compares the citation’s individual metadata fields against available evidence, identifies inconsistencies, repairs citation information where possible, and provides an auditable explanation of the resulting verdict.

AI Citation Verification

The central function of SourceVerify is citation verification. Users can paste individual references or provide larger collections of references for checking. The system determines whether the cited publication can be matched to an actual scholarly record.

Rather than simply assigning a probability that a citation is genuine, SourceVerify compares individual citation fields against evidence and reports what matched, what did not match, and what information was corrected.

Detecting Fabricated References

SourceVerify is designed to identify references that do not correspond to real publications. This is particularly relevant when working with AI-generated academic text, since a citation can look professionally formatted while containing an incorrect or nonexistent paper.

The platform’s verification process does not mark a reference as verified unless supporting evidence is found. References that cannot be confirmed may instead receive an Unverified result or be sent for human review.

Field-by-Field Verification

SourceVerify breaks a citation into individual components, including title, authors, publication year, venue, and identifier information such as a DOI. Each field can then be compared with information retrieved from relevant sources.

This approach helps distinguish between a completely fabricated citation and a genuine publication that contains smaller metadata errors, such as an incorrect publication year or misspelled author name.

Reference Error Repair

The platform does more than flag problematic references. It can identify mismatched or missing information and provide corrected citation details where supporting evidence is available.

Examples include correcting an author name, fixing a publication year, resolving an incorrect title, identifying a missing DOI, or repairing other bibliographic metadata.

Evidence and Traceability

Each verification result is designed to show the evidence behind the decision. SourceVerify reports which fields matched, which fields differed, what was repaired, and where the supporting evidence was found.

This creates a more transparent verification process than systems that provide only a numerical confidence score without showing how the conclusion was reached.

SVRIS Verification Standard

SourceVerify uses its SVRIS, or SourceVerify Reference Identity Standard, as an open standard for citation verification. The standard defines a deterministic process for comparing reference metadata and classifying verification results.

SVRIS assigns field-level labels such as MATCH, CONTAINS, ABSENT, UNCONFIRMED, and CONTRADICTION. These labels contribute to four primary outcomes: VERIFIED, VERIFIED WITH ERRORS, NEEDS HUMAN REVIEW, and UNVERIFIED.

Deterministic Verification

The SVRIS methodology is designed to avoid opaque probability-based decisions. The same citation information should follow the same verification rules, making the resulting classification reproducible and easier to audit.

This is particularly useful for academic and editorial workflows where users may need to understand why a reference was accepted, rejected, or flagged for additional review.

Academic Database Searching

SourceVerify gathers evidence from academic search and indexing sources. Its public documentation identifies OpenAlex and Google Scholar among the sources used during verification, while additional searches can include Google Search and the pages associated with a reference.

The platform states that its quick citation checker searches more than 300 million scholarly works, while full account-based verification can use broader source coverage.

Bulk Reference Verification

Researchers can verify complete reference lists rather than checking every citation individually. SourceVerify supports bulk processing of bibliographies and states that hundreds of references can be processed in a single pass.

Bulk verification can be useful for academic manuscripts, theses, dissertations, systematic reviews, editorial submissions, and other documents containing large reference sections.

PDF Reference Checking

SourceVerify can process uploaded PDF documents to extract references and verify them. Users can upload academic documents instead of manually copying each citation into the verification interface.

This workflow can be useful when a researcher already has a complete manuscript or paper and wants to conduct a reference-integrity check before submission.

Reference Formatting

SourceVerify includes reference formatting alongside verification and repair. Users can convert references into supported academic styles including APA, MLA, Chicago, and other formats.

Formatting is separate from verification: a reference can be formatted correctly even when its underlying publication information is inaccurate. SourceVerify therefore combines formatting with existence and metadata checks.

Verification Verdicts

SourceVerify provides four primary verification outcomes. Verified indicates that sufficient evidence supports the reference. Verified with Errors indicates that the underlying work exists but contains citation inaccuracies.

Needs Human Review is used when available evidence is ambiguous, while Unverified indicates that the system could not sufficiently confirm the reference. These classifications help users distinguish certainty from cases requiring additional judgment.

Multilingual Verification

SourceVerify supports verification across commonly used research languages. Its interface is currently available in English, French, Spanish, and Portuguese, with additional language support planned.

The platform’s verification system is intended to work with references beyond English-language academic literature, making it suitable for researchers working with international publications.

Browser Extensions

SourceVerify provides browser extensions for Chrome and Firefox. The extensions allow users to highlight or work with references directly from web pages and access verification without repeatedly switching between browser tabs.

The platform describes integrations with environments including ChatGPT, Google Docs, Overleaf, and arXiv, allowing citation checking to become part of an existing research workflow.

REST API

SourceVerify provides a REST API for developers who want to incorporate citation verification into their own applications or research systems.

The API accepts references programmatically and returns verification results in JSON format. Batch processing is also available for higher-volume workflows, with separate volume pricing for API usage.

CLI and Claude Code

SourceVerify provides a command-line interface and a dedicated skill for Claude Code. This allows developers and researchers to run reference verification from a terminal-based workflow.

The CLI can be used independently, while the Claude Code integration provides a way to verify references without leaving an AI-assisted coding or research environment.

MCP Integration

SourceVerify also provides an MCP server that can connect its verification functionality to compatible AI assistants. Supported workflows allow an assistant to submit references, retrieve verification results, check account balances, and access verification history.

The MCP connection is designed to avoid requiring users to manually copy an API key. After authentication and approval, compatible assistants can interact with SourceVerify’s verification functions.

Research and Academic Workflows

SourceVerify can be used at several stages of an academic workflow. Researchers can check citations while preparing a manuscript, verify an existing bibliography, repair incorrect metadata, or perform a final reference check before submission.

Editors, reviewers, and research organizations can also use bulk verification to examine large collections of references during peer review, publication, grant evaluation, or research due diligence.

AI-Generated Citations

One of SourceVerify’s primary use cases is checking citations produced by generative AI systems. AI-generated references can contain fabricated papers, incorrect author combinations, inaccurate dates, or other metadata errors even when the citation looks correctly formatted.

SourceVerify provides a separate verification step that checks the reference against external evidence rather than relying on the language model that generated it.

Research Integrity

Reference accuracy is an important part of academic research integrity. A properly formatted citation is not necessarily a valid citation, which is why SourceVerify separates formatting from verification.

The platform’s existence-first approach is intended to establish that a cited work can actually be identified before it is treated as a verified reference.

Pricing and Plans

SourceVerify offers a free tier and paid monthly subscriptions. The Free plan provides 25 verification tokens and 25 formatting tokens as a one-time allowance and does not require a credit card.

The Starter plan costs $20 per month and includes 125 verification tokens and 125 formatter tokens. The Workhorse plan costs $35 per month and includes 500 verification tokens and 500 formatter tokens.

The Powerhouse plan costs $65 per month and includes 1,500 verification tokens and 1,500 formatter tokens. Custom enterprise pricing is available for higher-volume requirements.

Who Is SourceVerify For?

SourceVerify is designed for researchers, students, academics, peer reviewers, editors, publishers, research organizations, and teams that need to validate references at scale.

It can also be useful for teams reviewing AI-generated content where citation accuracy needs to be checked before information is published or incorporated into formal research.

Reference Verification Before Publication

A practical workflow is to collect references during drafting, run verification on the complete bibliography, review flagged entries, repair incorrect metadata, and conduct a final verification before submission or publication.

This approach treats reference verification as a dedicated quality-control stage rather than assuming that correctly formatted references are automatically genuine.

Limitations

SourceVerify is a reference verification service rather than a system for determining whether the substantive claims in a research paper are scientifically true. Its terms explicitly state that the service does not verify the substantive truth of research claims.

Verification results should therefore be reviewed by the researcher or responsible professional, particularly when a citation receives a Needs Human Review result or when the source is difficult to index.

Frequently Asked Questions

What is SourceVerify?

SourceVerify is an AI-powered reference verification platform that checks academic citations for existence, metadata accuracy, errors, and supporting evidence.

Can SourceVerify detect fake citations?

Yes. SourceVerify is designed to identify references that cannot be matched to genuine published records and classify them as unverified when sufficient evidence cannot be found.

Can SourceVerify verify an entire bibliography?

Yes. Users can paste complete reference lists or upload PDFs containing references for bulk verification.

Does SourceVerify repair citation errors?

Yes. The platform can identify incorrect or missing citation details and provide corrected metadata when supporting evidence is available.

What citation styles does SourceVerify support?

SourceVerify supports APA, MLA, Chicago, and other academic citation styles for reference formatting.

Does SourceVerify check PDFs?

Yes. Users with an account can upload PDF documents and have the references extracted and checked in bulk.

What is SVRIS?

SVRIS stands for SourceVerify Reference Identity Standard. It is an open verification standard that defines deterministic rules for comparing citation fields and producing transparent verification outcomes.

Is SourceVerify free?

Yes. SourceVerify provides a free allowance of 25 verification tokens and 25 formatting tokens. Paid plans provide substantially larger monthly allowances.

How much does SourceVerify cost?

Paid subscriptions currently start at $20 per month for Starter. Workhorse costs $35 per month and Powerhouse costs $65 per month.

Does SourceVerify have an API?

Yes. SourceVerify provides a REST API for programmatic citation verification and batch processing.

Does SourceVerify support MCP?

Yes. SourceVerify provides an MCP server that can connect its verification functions to compatible AI assistants and development environments.

Does SourceVerify detect plagiarism?

No. SourceVerify’s Terms of Use specifically distinguish reference verification from plagiarism detection. The service focuses on checking whether citations exist and whether their bibliographic information is accurate.

Share Your Experience: