There is no single, universally accepted “master index” that rates the information quality and the accuracy of every website on the internet — nothing equivalent to a credit score for information. What exists instead is a patchwork of services built by journalists, librarians, academics, and search engines, each measuring a different slice of “trustworthiness”: some score entire publications, some score individual claims, some score the author rather than the outlet, and some are invisible algorithms you can only infer the existence of. Used together, they form a reasonably solid triangulation system. Used alone, any one of them can mislead you.
Investigators apply the same verification habits in OSINT techniques, cross-checking each claim against several independent public sources before trusting it. Open source intelligence techniques also rely on fact-checking sites like this one to confirm quotes, images and statistics.
A disclaimer and a warning before you rely on any of these:
Every tool below reflects the judgment, methodology, and — in some cases — the funding sources of the organization that built it. Bias-and-accuracy raters like MBFC and Ad Fontes are themselves regularly accused of bias by people across the political spectrum. Crowdsourced systems can be gamed or brigaded. Bibliometric scores (like Impact Factor or h-index) measure citation activity, not correctness — a frequently cited paper can still be wrong, and a correct paper in a niche journal may have a low score. Even Google’s E-E-A-T is not a transparent, auditable formula; it is inferred from patent filings, leaked documents, and Google’s own vague guidance.
Treat every score below as one data point, not a verdict. Cross-reference at least two independent tools — ideally from different categories — before deciding a source is reliable or unreliable, and never let an automated score substitute entirely for your own reading of the primary material.
For full clarity, even our selection below can be argued, so as our pros, cons and comments on each one! You are welcome to comment, suggest changes, so as other tools we forgot or were not aware of > here <.
In short, despite our “Cons” or comments, if we listed a tool/site here, we thought it worths checking if it fits your purpose.
Key Takeaways
- No single authority decides what’s true online: always cross-check at least two independent sources before trusting a claim.
- Outlet-level raters and claim-level fact-checkers do different jobs: a “credible” source can still publish one false claim, and vice versa.
- Citation metrics measure prestige, not accuracy: a highly cited paper can still be wrong or later retracted.
- Paper mills and fraud have surged in academic publishing: always check a paper against retraction databases before relying on it.
- Virality is not a truth signal: content spreads based on emotional impact, not factual accuracy.
- Every rating tool reflects its creator’s methodology and biases: treat scores as one data point, not a verdict.
- Transparency is the strongest trust signal: favor sources that disclose funding, authorship, and correction policies.
- Context determines reliability: the same source can be trustworthy for one claim and unreliable for another.
- Manual judgment still matters: frameworks like CRAAP remain essential when no database covers a source.
- Slowing down beats any tool: pausing before sharing prevents more misinformation than any single fact-checker.
- Bonus basic rule of thumb: do no trust blindly what a site is claiming about itself (“we are the #1 …”).
Tip: you may find directly some of these tools in our > Online Tools Database <, either in the “Network” section or in the “OSINT” section.
Outlet-Level Credibility Raters
These tools evaluate an entire website or publication — its track record, ownership, and editorial practices — rather than any single article:
NewsGuardA commercial rating service that employs trained journalists (not algorithms) to evaluate the more than 35,000 news sources responsible for approximately 95% of all the news and information consumed and shared online across the U.S., U.K., Canada, Germany, France, Italy, Austria, Australia, and New Zealand. Each site is scored 0–100 based on nine apolitical criteria, weighted differently, that together determine the score, covering things like whether the outlet repeatedly publishes false or egregiously misleading content, corrects errors, and discloses ownership. The result is displayed as a “Nutrition Label” — a green, red, satire, or platform rating shown via a shield icon next to links on search engines and social media when the browser extension is installed. Before publishing a negative rating, NewsGuard’s analysts contact the publisher, seek comment, and give them a chance to resolve the issue or respond, and that response is published alongside the rating. Particularities: It explicitly separates satire from “fake news” — a humor or satire site is given a separate designation rather than scored on the nine criteria.
Best for: a quick, browser-level trust signal while reading news day-to-day, and for understanding who owns and funds a site. | Media Bias/Fact Check (MBFC)An independent, free, searchable database now covering over 10,000 media sources, journalists, politicians, and countries. Rates each outlet on two separate axes — political bias (left to right) and factual accuracy (very high to low) — and maintains classification categories including Fake News, Satire, Extreme Bias, Conspiracy Theory, Rumor Mill, State News, Junk Science, Hate News, Clickbait, Proceed with Caution, Political, and Credible. Particularities: It’s run by a small team rather than a large newsroom, and its write-ups for each outlet typically cite specific failed fact-checks as evidence.
Best for: a fast gut-check on whether an unfamiliar outlet has a track record of distorting facts to fit a political agenda. |
Ad Fontes Media (Media Bias Chart)A visual chart, not in the search results above but well known publicly, that plots outlets on two axes — reliability (vertical) and political lean (horizontal) — using panels of analysts from different political perspectives reviewing the same articles.
Best for: visually comparing many outlets at once rather than deep-diving on one. |
Claim-Level Fact-Checkers
These don’t rate a website — they rate a specific statement, video, or viral post.
PolitiFactA fact-checking website that rates the accuracy of claims by elected officials and others, run by editors and reporters from the Tampa Bay Times. Each claim is placed on the “Truth-O-Meter,” a scale running from True to False, with the most ridiculous falsehoods earning the lowest rating, “Pants on Fire”. Particularities: it focuses specifically on political speech — debates, ads, interviews, social posts by public figures — rather than general internet claims.
Best for: checking a specific quote or political claim you’ve seen circulating. | SnopesThe oldest and largest general fact-checking site on the internet, founded in 1994, originally focused on urban legends and now covering viral claims of all kinds. Claims are rated on a granular scale: ratings indicate whether the primary elements of a claim are demonstrably true, true but with a misleading ancillary detail, a genuine mix of true and false, false but rooted in some truth, or simply demonstrably false.
Best for: checking whether a viral photo, video, or “did you hear about…” story is real. |
FactCheck.org, AFP Fact Check, Reuters Fact Check, Lead StoriesA cluster of similar claim-checking operations, several run by major wire services. Lead Stories in particular uses its “Trendolizer” technology to identify trending content, which is then fact-checked by a team of journalists. Reuters Fact Check is a signatory to the International Fact-Checking Network’s code of principles, as is AFP’s fact-checking arm.
Best for: breaking-news-adjacent claims, especially ones spreading internationally. | https://www.afp.com/en/our-offer/afp-fact-check |
To be Noted: fact-checking organizations have faced a serious funding squeeze in the last 3 years, with the International Fact-Checking Network reporting that 45.3% of accredited fact-checkers saw revenue declines in 2025. Several major funding sources dried up around the same time: Meta ended its U.S. fact-checking partnership, USAID was shut down after a government grant freeze, and the National Endowment for Democracy suspended related grants. This left many organizations cutting staff and relying more heavily on a smaller pool of grants, even as public interest in their fact-checks kept growing.
Standards Bodies (Raters of the Raters)
These organizations don’t rate content directly — they certify fact-checking organizations themselves, which is useful when you’re trying to judge whether to trust a fact-checker you’ve never heard of.
International Fact-Checking Network (IFCN)The IFCN reviews fact-checkers for compliance with its code of principles and issues a certification, valid for one year, to publishers who pass the audit; it listed around 170 member organizations as of mid-2024. Best for: checking whether an unfamiliar “fact-check” site is a credentialed operation or a one-person blog borrowing the format. | European Fact-Checking Standards Network (EFCSN)Launched in 2023 following a European Commission project, it reviews fact-checkers for compliance with a code of standards and certifies those who pass review by two independent assessors. The code requires that a certified fact-checker reviews statements from all parties and sides, examines discrete claims, transparently identifies sources and methods, discloses funding and affiliations, and treats news and information as its primary mission. Best for: the European equivalent of the IFCN check — particularly relevant for EU-based outlets. |
The Trust Project (Trust Indicators)An international consortium of news organizations that developed a shared set of “Trust Indicators” — standardized disclosures about funding, authorship, and correction policy — that participating outlets display, sometimes marked with a “Trust Mark.” Unlike NewsGuard or MBFC, it doesn’t produce a score. It’s closer to a transparency certification: does the outlet show its work?
Best for: verifying that a specific outlet is accountable about its funding and editorial process, when the indicators are present. |
Scientific and Academic Indices
For technical or scientific claims, none of the above tools are designed to help — academic publishing has its own, much older rating infrastructure, built around citation, not journalistic review.
SJR (SCImago Journal Rank)Ranks academic journals by prestige, based on both the number of citations a journal receives and the weight of the journals citing it (a citation from a top-tier journal counts more than one from an obscure one — similar in spirit to Google’s original PageRank).
Best for: a first filter when you land on an unfamiliar journal name and want to know if it’s a recognized venue or a “predatory” pay-to-publish outlet. | h-indexMeasures an individual researcher’s combined productivity and citation impact (an h-index of 20 means the researcher has at least 20 papers that have each been cited at least 20 times).
Best for: checking whether someone presented as an authority has a genuine, substantial publication record in that specific area — not just any field. |
Impact Factor (IF)A journal-level metric: the average number of times articles published in a journal over the prior two years were cited in the current year.
Best for: comparing journals within the same field — never across fields, and never as a substitute for actually reading the paper. |
A note on recent problems with academic indices:
In recent years, bibliometric tools like Impact Factor, SJR, and h-index have come under serious strain from industrialized research fraud. “Paper mills” — commercial operations that mass-produce fake or plagiarized studies, fabricate data, and sell co-authorship slots — have flooded some journals with fraudulent submissions, and citation cartels (researchers or brokers who arrange mutual citations purely to inflate metrics) have further distorted these scores. Publishers including Hindawi, Springer Nature, and others have retracted thousands of papers in mass purges since 2022–2023 after discovering compromised special issues and fabricated peer review. While these schemes have been documented across many countries — including well-publicized cases in Russia, Iran, the US, and parts of Europe — research-integrity watchdogs like Retraction Watch have noted that paper-mill output has been disproportionately concentrated in certain fast-growing academic markets, largely driven by intense institutional pressure to publish for hiring, promotion, and funding rather than by any characteristic of individual researchers. The practical upshot is that a high Impact Factor or SJR may (you to evaluate) no longer guarantees a journal’s special issues or recent volumes are clean — checking a journal or paper against retraction databases (see tools in next chapter below) alongside its citation metrics is now a necessary extra step.
Post-Publication Integrity Tools (Catching Fraud After the Fact)
Every tool in previous section — SJR, h-index, Impact Factor — measures prestige through citation. None of them were built to catch fraud, and in recent years a separate category of tools has emerged specifically to do that: finding papers that were fabricated, plagiarized, peer-review-rigged, or bought outright, often well after publication:
Retraction Watch DatabaseA free, public database — now containing over 64,000 retraction records — that tracks retractions, expressions of concern, and corrections across academic publishing. Retraction Watch acquired the database from the Center for Scientific Integrity in 2023 and now updates it daily; the dataset is also distributed through Crossref. Entries are gathered from publisher websites, scholarly databases, web searches, and tips from the research community, then categorized by reason for retraction (plagiarism, data fabrication, image manipulation, compromised peer review, paper-mill origin, and dozens of other tags). Particularities: it’s the only tool here aimed squarely at individual papers, after the fact, rather than journals or authors in the abstract. It also tracks adjacent integrity signals most other tools ignore entirely — the site maintains a running list of retractions specifically tied to COVID-19 misinformation and a list of mass editorial-board resignations.
Best for: checking whether a specific paper, author, or journal has a documented retraction history before you cite it — genuinely a “did this already get caught” check. | PubPeerA post-publication peer review platform, launched in 2012, where anyone — researchers, students, or the original authors — can comment on a published paper’s validity, methodology, data, or images, anonymously or under their own name. Unlike formal peer review, which happens privately before publication, PubPeer comments are public and happen after the fact, often flagging image duplication, statistical anomalies, or data inconsistencies that formal review missed. These public threads have prompted numerous real-world retractions and several high-profile misconduct investigations. Particularities: Comments are moderated but participation is open to all, including anonymous “science sleuths” who specialize in spotting manipulated images or duplicated data across papers.
Best for: searching a specific paper or author to see if independent reviewers have already raised unresolved concerns, even if no formal retraction has happened yet. |
Problematic Paper Screener (PPS)A free, automated tool built by computer scientists Guillaume Cabanac and Cyril Labbé that continuously scans roughly 130–140 million indexed scientific publications for telltale signs of fabrication. Its best-known detector hunts for “tortured phrases” — garbled synonym substitutions produced when fraudsters run stolen text through paraphrasing software to dodge plagiarism checkers, producing absurd phrases like “fake neural organization” instead of “artificial neural network.” Nine other detectors look for different red flags, including nonsense papers generated by the spoof program SCIgen, leftover AI chatbot artifacts such as “as an AI language model, I cannot…”, and papers that cite a suspicious number of already-retracted references. Particularities: It has flagged nearly 19,000 papers containing multiple tortured phrases and over 280 entirely computer-generated gibberish papers, some of which were still circulating in the literature at time of detection. The tool’s methodology was significant enough that Nature named it one of the year’s key developments in science, and Springer Nature has since built its own in-house detector directly on top of the PPS’s public tortured-phrases catalogue.
Best for: technically minded users and journalists who want to check whether a paper’s text shows known fingerprints of paraphrasing fraud or AI-generated filler — and for publishers, as a pre-publication screen. | STM Integrity Hub and Publisher-Side Detection ToolsA shared, cloud-based initiative run by the International Association of Scientific, Technical & Medical Publishers (STM), giving member publishers a common environment to screen submitted manuscripts for integrity issues and to access third-party detection tools, including donated tools like Springer Nature’s AI-text detector. This represents a shift from individual publishers fighting fraud in isolation toward cross-industry, shared detection — alongside parallel efforts like the United2Act coalition.
Best for: understanding what’s happening upstream, at the publisher level, before a flawed paper ever reaches a database like Retraction Watch’s. https://stm-assoc.org/what-we-do/strategic-areas/research-integrity/integrity-hub/ |
Search Engine and Algorithmic Frameworks
Google’s E-E-A-T Score
An internal framework — not a visible public score — that Google’s Search Quality Rater Guidelines use to train and evaluate the ranking algorithm. You can’t look up a site’s “E-E-A-T score” anywhere; you can only infer its effects from how content ranks. Its four components:
- Experience — has the author actually done or used the thing they’re writing about?
- Expertise — does the author have relevant credentials or demonstrated knowledge?
- Authoritativeness — is the site recognized as a go-to reference on this topic?
- Trustworthiness — is the site secure, transparent about authorship, and accurate?
This is one of the reason why, for a technical topic, Google’s results tend to surface .gov, .edu, and established .org domains above newer commercial sites — those domains structurally score higher on authoritativeness and trust signals.
- Pros: operates at internet scale, continuously, on virtually everything indexed — no other tool comes close in coverage.
- Cons: completely opaque. There is no published formula, no appeals process, and no way to audit a specific ranking decision. It is also a ranking signal, optimized to predict what searchers will find satisfying — not a pure accuracy measure, and the two can diverge (highly satisfying content isn’t always the most correct content).
Best for: understanding why certain domains consistently outrank others for technical and medical topics — useful context, but not a tool you can query directly.
Manual Evaluation Frameworks
Unlike everything above, these aren’t databases you search — they’re checklists you apply yourself to any source.
The CRAAP Test
Developed by librarians at California State University, this is the standard taught in most university research-methods courses:
- Currency — how recent is the information, and does that matter for this topic?
- Relevance — does it actually address what you need?
- Authority — who wrote or published it, and what are their credentials?
- Accuracy — is it verifiable, well-sourced, and free of errors you can check?
- Purpose — why does this content exist — to inform, persuade, entertain, or sell?
Pros: free, universally applicable to any source type (website, book, video, social post); forces deliberate, structured thinking rather than a gut reaction. Promotes self-evaluation good practices.
Cons: entirely manual — there’s no score or database, just better personal judgment; quality of the result depends on how rigorously you apply it.
Best for: the foundation of any online info search. Evaluating a source with no entry in any of the databases above — which, for most niche or new websites, will be the common case.
Crowd-Governed Reference Lists
Wikipedia’s Perennial Sources List (WP:RSP)
A community-maintained list on English Wikipedia, established in 2018, that classifies sources by degrees of reliability based on accumulated editorial discussion.
The list is not an independent document — it’s derived from the conclusions of formal Requests for Comment (RfCs) and other discussions, and is intended as a summary of community consensus. Sources fall into tiers including generally reliable, “marginally reliable” (no consensus, unclear, or additional considerations apply) — examples cited include National Review, Jezebel, and Salon.com — and “generally unreliable,” meaning a source “should normally not be used” and “should never be used for information about a living person”. The most problematic sources, including certain user-generated content sites like Amazon customer reviews, Discogs, and TV Tropes, can be formally deprecated.
Particularities: context matters enormously — a source’s reliability depends on the specific claim it’s being used for, and the list explicitly warns that when in doubt, readers should consult the underlying linked discussions rather than just the summary icon. Consensus can change over time, and the list is updated to reflect newer discussions that consider new evidence.
- Pros: free, extensive (hundreds of sources), transparently sourced to specific debates rather than one editor’s opinion, and continuously updated; the underlying discussions are public, so you can read the actual reasoning rather than trust a black-box score.
- Cons: it rates reliability for Wikipedia’s specific sourcing purposes — a source can be “generally unreliable” by Wikipedia’s encyclopedic standards while still being useful for other purposes (e.g., a primary-source company blog is “unreliable” for neutral facts about a competitor but perfectly fine as a source for the company’s own stated positions). It’s also occasionally criticized as reflecting the political composition of Wikipedia’s editor base rather than a neutral standard.
Best for: checking whether a specific outlet has a documented history of disputes over accuracy, and reading the actual arguments for and against it — genuinely one of the more transparent tools in this whole list, precisely because you can see the underlying debate, not just the verdict.
https://en.wikipedia.org/wiki/Wikipedia:Reliable_sources/Perennial_sources
Motre a Global Tip Than a Conclusion
No tool above is sufficient on its own, and they’re strongest in combination across categories:
- For a news outlet: check NewsGuard or MBFC for the outlet’s track record, then check whether Wikipedia’s RSP has a documented consensus on it.
- For a viral claim: check Snopes or PolitiFact directly, then verify the claim’s underlying source rather than just trusting the fact-checker’s headline.
- For a technical or scientific claim: look at the journal’s SJR and the author’s h-index as a baseline filter, but actually read the paper’s methodology — citation metrics don’t replace that.
- For anything with no entry anywhere: fall back to the CRAAP test and do the manual work yourself.
The common thread across every credible tool here — NewsGuard’s right-of-reply, the IFCN’s annual re-audits, Wikipedia’s public discussion trail — is that they show their reasoning rather than just asserting a verdict. That transparency, more than the specific score, is the real signal of a tool worth trusting.
FAQ
What’s the single fastest way to check if a news story is true?
Search the claim’s key facts plus the word “fact check” and see if PolitiFact, Snopes, or Reuters Fact Check has already covered it. If no fact-checker has touched it yet, trace the story back to its original source rather than trusting whoever shared it.
Why can’t I just trust a website because it looks professional?
Polished design is cheap to fake and tells you nothing about editorial standards. Check who owns the site, whether authors have real credentials, and whether claims are sourced — not how the site looks.
Is a high citation count proof that a scientific paper is correct?
No. Citation counts measure influence and visibility, not accuracy, and fraudulent or later-retracted papers can rack up citations for years before anyone catches the problem. Always check whether a paper has been retracted before relying on it.
What’s the difference between misinformation and disinformation?
Misinformation is false information shared without knowing it’s false, often by accident. Disinformation is false information created and spread deliberately to deceive, usually for political or financial gain.
Can I trust an article just because it cites scientific studies?
Not automatically — check whether the cited study actually says what the article claims, since headlines often overstate or distort findings. Also check if the study was peer-reviewed, replicated, or later retracted.
How do I know if an image or video is real?
Reverse image search the photo or a key frame of the video to see where and when it first appeared online. Manipulated and out-of-context media is one of the most common ways false claims spread, so treat dramatic visuals with extra skepticism.
Are Wikipedia articles reliable?
Wikipedia itself is generally a good starting point for context, but its true value for fact-checking is the sources cited in the footnotes. Check those underlying sources directly rather than treating the Wikipedia text itself as the final word.
Why do different fact-checkers sometimes disagree?
Fact-checking involves human judgment, especially on nuanced or politically charged claims, so reasonable disagreement happens. When sources conflict, check more than one fact-checker and look at the original evidence yourself rather than picking whichever verdict you prefer.
What does it mean if a source is “peer-reviewed”?
It means other experts in the field reviewed the work before publication and judged the methodology sound enough to publish. It is not a guarantee of correctness — peer-reviewed papers are still retracted regularly, especially as paper-mill fraud has grown in recent years.
Shold I trust a claim just because it has lots of likes or shares?
No — virality measures how emotionally engaging or shareable content is, not how true it is. False claims often spread faster than corrections precisely because they’re designed to provoke a strong reaction.
How can I tell if a website has a political agenda?
Check an independent bias rating from a tool like Media Bias/Fact Check or Ad Fontes Media, and look at who funds the site and who owns it. A site can still report facts accurately while consistently choosing which facts to emphasize, so bias and inaccuracy are not the same problem.
What’s one habit that improves my judgment of online information the most?
Slow down before sharing — pause and ask who benefits if you believe this, and check the original source before reacting. Most misinformation succeeds because people share on impulse rather than because the claim is convincing once you actually check it.
