ArXiv – Bias and Credibility

ArXiv is rated Least biased with Mixed factual reporting by Media Bias Fact Check.

Detailed Report

Bias Rating: LEAST BIASED (-0.5)
Factual Reporting: MIXED (4.6)
Country: USA
MBFC’s Country Freedom Rating: MOSTLY FREE
Media Type: Website
Traffic/Popularity: High Traffic
MBFC Credibility Rating: MEDIUM CREDIBILITY

History

arXiv is an open-access research-sharing platform and preprint archive covering physics, mathematics, computer science, quantitative biology, quantitative finance, statistics, electrical engineering and systems science, and economics. According to its About page, arXiv was founded by Paul Ginsparg in 1991 and, after decades of being hosted by Cornell University, became an independent nonprofit organization in 2026.

arXiv hosts more than three million scholarly articles and provides submission, retrieval, search, distribution, API access, curation, and preservation services. Importantly, arXiv states that submissions are moderated but not peer-reviewed, and that the contents of submissions are the responsibility of the authors and are presented “as is” without warranty or approval of the assumptions, methods, results, or conclusions.

Read our profile on the United States government and media.

Funded by / Ownership

arXiv is an independent nonprofit organization governed by a Board of Directors. Day-to-day operations are led by a Chief Executive Officer, staff, volunteer subject moderators, and advisory councils.

According to its Funding Support page, arXiv is funded by Simons Foundation International, member institutions, affiliates, sponsors, foundations, donors, and individuals. Its Membership page states that member institutions contribute more than 50% of arXiv’s operating budget. Membership rates range from $1,000 to $10,000 depending on institutional tier, while affiliates and sponsors may contribute larger amounts.

Analysis / Bias

arXiv is rated Least Biased because it is a broad research repository rather than a political or ideological publication. It hosts submissions across many scientific and technical disciplines and does not appear to promote a consistent political viewpoint. Its mission is open research distribution, not editorial commentary.

However, arXiv is not rated Pro-Science because it does not peer review or independently verify the scientific accuracy of hosted papers. Its moderation policy says submissions are checked for topical relevance, scholarly value, format, and basic standards, but moderation is not peer review. arXiv may decline or remove papers, but it does not validate findings before publication.

Some arXiv submissions are legitimate scholarly works, such as Morphing Graphs on Hyperbolic Surfaces, a technical mathematics and computational geometry preprint. arXiv is widely used by researchers to share early results before journal or conference peer review.

At the same time, weak or unreliable work can appear on the platform. For example, WithdrarXiv: A Large-Scale Dataset for Retraction Study documents more than 14,000 withdrawn arXiv papers and categorizes the stated reasons for withdrawal. Withdrawals do not necessarily indicate misconduct or misinformation, but they demonstrate that papers containing errors, unresolved concerns, or other significant problems can be publicly distributed before those issues are identified. The Retraction Watch Database also documents arXiv-related cases involving data concerns, authorship disputes, institutional approval problems, errors, bias concerns, and computer-generated content.

Further, Science reported that arXiv has faced a rising volume of poor-quality and AI-generated submissions, leading the platform to require first-time submitters to receive endorsement from established arXiv authors. According to the article, arXiv’s rejection rate rose from about 4% historically to an estimated 10% to 12% amid the increase in low-quality AI-generated submissions.

Readers should also understand the limits of preprints. Snopes explains that preprint research has not been peer-reviewed and has not yet been rigorously vetted for limitations, errors, or methodological problems. This is especially important because arXiv papers are often cited online as though they are established scientific findings, even when they are preliminary, disputed, or later withdrawn.

Overall, arXiv is an important open-access scholarly archive, but it should not be treated as a peer-reviewed scientific publisher. Its moderation system provides meaningful quality control, but flawed analysis, unsupported conclusions, serious errors, and inadequately reviewed AI-generated content can still appear. Because arXiv does not independently verify the accuracy of submissions before making them publicly available, we rate its factual reporting Mixed.

Failed Fact Checks

• None found for arXiv as an organization by IFCN-approved fact-checkers.
• Individual arXiv preprints have been withdrawn or retracted for errors, data concerns, authorship issues, institutional approval problems, bias concerns, and computer-generated content, as documented by WithdrarXiv and the Retraction Watch Database.

Overall, we rate arXiv Least Biased and Mixed for factual reporting. It is a politically neutral academic preprint repository, but submissions are not peer-reviewed, and arXiv does not independently verify their methods, evidence, or conclusions. While most papers are legitimate scholarly works, flawed, unsupported, or AI-generated material can appear. Readers should treat arXiv papers as preliminary until independently reviewed or published in a peer-reviewed venue. (D. Van Zandt 06/25/2026)

Source: https://arxiv.org/

Last Updated on July 25, 2026 by Media Bias Fact Check

Left vs. Right Bias: How we rate the bias of media sources