2026-04-21: Was Snopes.com making silent updates to its articles before 2021? [Rating: TRUE]

Figure 1: Snopes.com has no public captures on the Internet Archive’s Wayback Machine prior to October 2021.

Snopes.com is a well-known fact checking website. It has been rated trustworthy by multiple rating organizations [1]. However, in 2021, BuzzFeed broke a story that, in fact, Snopes was plagued by plagiarized articles, and that authors were being told to plagiarize and then silently make updates to their articles after publication in the name of SEO.

One of the byproducts of the BuzzFeed investigation is that there are no publicly available captures of Snopes.com in the Wayback Machine prior to 2021. Snopes’s official statement regarding this is that the founder, who was named responsible for the plagiarism, had a policy against Wayback captures; but, now that he was removed from duty, the board was going to change the policy going forwards. In fact, having no captures before 2021 means that the archives available give a more trustworthy historical view of Snopes.com than exists in reality. This kind of manipulation is a form of data craft that misuses web archives [2]. Regardless, users do not have a way to access snopes.com captures before 2021 in the Wayback Machine.

Finding archived copies of Snopes.com before 2021

The Wayback Machine is not the only web archive. In addition, the Wayback Machine accepts donated crawls from other organizations. One of these organizations is Common Crawl, which started crawling in 2008. It seems almost certain that Common Crawl would include snopes.com captures, though probably not with enough frequency to show the silent edits on individual article pages. Searching in a random month (July 2016) within the plagiarism window shows that there are over 12,000 captures of snopes.com pages and resources during that month. Only 453 of these pages were archived a single time during this month, so future analysis with this approach may be possible.

Figure 2: In July 2016, there were over 12,000 captures of snopes.com pages via Common Crawl, with most having 2 or more captures. Source: index.commoncrawl.org

Another organization that has captures of snopes.com before 2021 is Archive-It. Specifically, one of the very first public captures of Snopes.com in the Wayback Machine belongs to an Archive-It organization which turns out to be Mark Graham, the Director of the Wayback Machine. Mark’s personal collections have been instrumental in analyses about news websites [3] and websites in other countries [4]. He has over 300 Snopes articles archived with frequency starting around 2018 in his collection.

We also used MemGator to investigate if there are pre-2021 captures of snopes.com in additional archives; we found that there are over 13,000 captures of the main page pre-2021 in archives including the Icelandic Web Archive, the Australian Web Archive, Archive Today, the Portuguese Web Archive, the Government of Canada Web Archive, along with Archive-It. Over 12,000 of the captures are on Archive-It.

Looking for Silent Updates

Previously, we found silent updates occurring on federal websites in February 2025 [5]. Other researchers also found silent updates on news websites in 2022 correlated with political bias of the news organization [6]. We are also researching how websites announce their changes and how trustworthy the announcements are [7]. Since Snopes.com is regarded as a trustworthy website, we would not have expected silent updates prior to reading the BuzzFeed article.

First, let’s take a moment to observe the current change presentation of Snopes.com. Snopes.com includes a last updated date, and a list of all changes. This would mean that the live web version of Snopes is presenting itself as “Updates summary,” the second most trustworthy level, according to the change presentation gradient [7].

Figure 3: Example of updates summary presentation on a snopes.com article.

However, using Mark Graham’s captures, we have found a number of pages with silent updates.

Example 1: Missing last updated information

The news story “White House Press Secretary Blasted for Sharing Infowars Video to Bar Reporter” states it was published on November 8, 2018. However, a quote from the article states, “On 11 November 2018, White House Counselor Kellyanne Conway admitted…” Based on the published date of November 8, this doesn’t seem possible. Examining a capture from Archive-It, this article used to include a manual last-updated date of November 12. It doesn’t appear Snopes included an updates summary in 2018 like it does in 2026. Sometimes, examining the machine-readable headers gives additional information; in this case, however, the x-archive-orig-last-modified header is the same as the capture time. The Snopes sitemap for that month indicates the page has been updated as recently as March 2025.

Figure 4: This snopes.com article originally stated it was updated, as shown above in an Archive-It capture. On the live web, as shown below, there is no mention of an update on this article anymore.

Example 2: Removal of AP articles

Snopes.com used to include exact text versions of Associated Press articles. Now, however, they redirect to the actual AP article. Given the plagiarism backhistory, this is not surprising. A capture on Archive-It shows the direct text of an AP article on snopes.com. The article link on snopes.com today redirects to the AP article rather than duplicating the text.

Example 3: Silent Update - Word-switching

The most interesting silent update we found was an article where many words had been replaced with synonyms after the fact. The article, “Did Obama Admin Build Cages That House Immigrant Children at U.S.-Mexico Border?” was published July 2, 2019 and includes no update date. However, we examined the capture from July 30, and it contained completely different text. Searching the web for this original text shows multiple citations. However, by August 9, the text changes to the version that matches the live web.

Figure 5: Differences between the July 2, 2019 and live web snopes.com article “Did Obama Admin Build Cages That House Immigrant Children at U.S.-Mexico Border?” show that the article was edited with a word-switching pattern. 

Example 4: Updates declaring SEO

David Mikkelson’s 2007 article on Mr. Rogers had an SEO update in 2022, which was disclosed in its updates summary. The original title is widely referenced, as determined by searching the web. If you search the web for “Updated SEO” on snopes.com, you can find other articles where they have noted updating the title as well. One of the reasons why Snopes has been rated more trustworthy and less biased by the organization Media Bias / Fact Check is because its titles are questions, which is interpreted as more neutral than a title that supports one conclusion over another. 

Figure 6: Google search for “Updated SEO” on snopes.com showing multiple articles with titles changed.

Outlook

Originally, we thought that a trustworthy website like Snopes.com would not have silent updates. However, after reading the BuzzFeed article about the rampant plagiarism on the site, we decided to investigate. We found evidence of silent updates using web archives beyond the Wayback Machine. This reduces Snopes.com’s trustworthiness level. Future work could determine if their plagiarism problem has truly ended, or if silent updates are still occurring post-2021.

References

[1] Yang et al. Are Fact-Checking Tools Reliable? An Evaluation of Google Fact Check. https://arxiv.org/abs/2402.13244v1

[2] Acker et al. The weaponization of web archives: Data craft and COVID-19 publics. https://misinforeview.hks.harvard.edu/article/the-weaponization-of-web-archives-data-craft-and-covid-19-publics/ 

[3] Weigle et al. Right HTML, Wrong JSON: Challenges in Replaying Archived Webpages Built with Client-Side Rendering. https://doi.org/10.1109/JCDL57899.2023.00022

[4] Ben-David et al. The Internet Archive and the socio-technical construction of historical facts.  https://doi.org/10.1080/24701475.2018.1455412

[5] Frew et al. Coming Back Differently: An Exploratory Case Study of Near Death Experiences of Webpages. https://digitalcommons.odu.edu/computerscience_fac_pubs/404/ 

[6] Tsoukaladelis et al. The Times They Are A-Changin’: Characterizing Post-Publication Changes to Online News. https://doi.org/10.1109/SP54263.2024.00033

[7] Lesley Frew. Detecting and reconstructing trustworthy edit histories using web archives. https://ws-dl.blogspot.com/2026/02/2026-02-03-detecting-and-reconstructing.html 

--Lesley