School districts are required to keep students from seeing obscene or harmful images online if they want federally subsidized internet access. An investigation by The Markup found that schools routinely go far beyond that, keeping students from information they need to complete their assignments and sometimes blocking access to information about abortion, sex education, and even suicide prevention resources for LGBTQ+ teens.

This censorship is possible because of software from technology companies that sorts much of the internet into categories. Districts decide for themselves what categories to block or allow, often creating custom categories to allow sites that would otherwise be blocked.

To identify what content students in U.S. schools are kept from seeing online, The Markup requested records of blocked websites from 26 school districts that spanned 14 states.

We started with the 15 districts that banned the most books during the 2022–23 school year, according to PEN America’s Index of School Book Bans because we wanted to see how the districts that were most frequently banning books were blocking the internet. When a district using iboss as its web-filtering software shared with us a straightforward report of blocked websites by category, we requested those reports from eight additional districts that used the same software. We identified those districts using GovSpend, a company that maintains a database of government contracts. We added two additional districts because they were attended by students we interviewed for the story. Another district, U-46 in Illinois, was selected as a trial run because it was previously covered by our investigative reporter.

For most districts, The Markup requested a list of websites blocked by web-filtering software in the last 30 days, including, for each website, its name, how the software categorized it, and the number of times users had attempted to access it. We also asked for a list of the websites the district manually blocked or unblocked, any policies that explain the criteria and process for blocking websites or challenging blocks, and the district’s contract with its web-filtering company.

Eleven districts denied the request for specific websites on cybersecurity grounds. The Markup does not believe these records are exempt from public records law and encouraged districts to redact any URLs they considered a security risk and release the rest. Two did so, but the first nine districts listed below did not comply. Another district refused to provide the records on other grounds.

Sixteen school districts provided The Markup with records, though the level of specificity varied.

After reviewing the records each district sent The Markup, we focused our reporting on blocked websites that could be important for students’ education, health, and safety. We found that these blocks included websites with sex education content, information about abortions, and resources for LGBTQ+ teens, the latter of which were blocked via categories named “human sexuality” or “sexual content.” These filters also blocked websites for basic educational research like Wikipedia. The districts that we examined much more frequently stopped students from browsing social media and playing games than they stopped anyone from watching porn.

Although overblocking was ubiquitous across The Markup’s sample, each district approached blocking differently, highlighting the fact that the internet can look very different to students in different parts of the country, something that drives equity concerns among opponents of these filters.

Across the 13 districts that released records of blocked websites and their category labels, only the Rockwood School District in Missouri had a specific category for LGBTQ+ websites.

Rockwood, which uses the ContentKeeper filter, keeps students at all grade levels from sites the software labels “human sexuality.” Almost 50 websites showed up as blocked via this category, including advocacy websites, dating websites, entertainment news, and resources all geared toward the LGBTQ+ community. None of them hosted pornographic images.

A number of LGBTQ+ websites show up blocked with a “High School Allowed” label, meaning they were blocked for younger students and/or for network guests, but not for high school students. Among those websites was itgetsbetter.org, glaad.org, and thetrevorproject.org.

The Securly filter, which Seattle Public Schools uses, has a “sexual content” category made up of sexual health and LGBTQ+ advocacy websites, but records from March showed the district leaves it unblocked.

Websites labeled “sex education” or “abortion” were blocked by the Beaufort, Granite, Houston, Jefferson Parish, Manatee County, Rockwood, Texarkana, and Washington County school districts. 

Websites under a “health” or “health sites” category were also blocked by the Beaufort, Fort Worth, Houston, Jefferson Parish, Palm Beach County, Rockwood, Seattle, Washington County, and Wentzville school districts.

Sites labeled “social media,” “social networking,” or “friendship” were blocked by 12 school districts: Beaufort, Fort Worth, Granite, Houston, Jefferson Parish, Manatee County, Palm Beach County, Rockwood, Seattle, Texarkana, Washington County, and Wentzville. The “friendship” category is defined as “sites that contain platonic friendship-related materials and social networking sites.”

Several districts blocked specific social media sites; Manatee, Wentzville, and Palm Beach blocked Twitter, Facebook, and Pinterest. Palm Beach also blocked TikTok, Snapchat, and Reddit.

All 13 districts blocked some content categorized under “porn” and/or “nudity,” though these categories did not show up in any district’s top 10 most-blocked categories. Only Palm Beach and Seattle had porn in the top 20.

The types of sites blocked by school filters varied by district, but entertainment websites dominated the top three blocked categories, whether they were related to social media (among the top three categories blocked in Manatee, Palm Beach, Fort Worth, and Washington County); games (among the top in Forsyth and Seattle); shopping (among the top in Palm Beach, Rockwood, and Washington County); or radio and TV streaming (among the top in Granite, Beaufort, and Fort Worth). Ads showed up in the top three in Jefferson Parish, Wentzville, and Houston. And catchall categories for websites the filter couldn’t label or marked “non-managed” also made the top three in Manatee, Rockwood, Texarkana, and Wentzville.

When school districts set up their filters, they create various categories of users that usually include students, staffers, and guests. Some districts also have different blocks for elementary, middle, and high schoolers. The records we obtained do not specify which type of user tried to access each blocked site.

Every student interviewed for this story said encountering blocked websites was a routine event when using school Wi-Fi. Still, in response to questions about why certain legitimate websites would be blocked (including cdc.gov, merriam-webster.com, stopbullying.gov, and autismspeaks.org), IT leaders in both the Rockwood and Manatee County school districts explained that they were only blocked under their respective filters’ more restrictive settings for guest users. These sites likely appeared in the data when adults attending district events or students visiting from other schools for various competitions tried to access the internet. One student specifically mentioned running into more aggressive blocking while trying to do homework during debate tournaments at other schools in his region.

For our investigation, we confirmed which sites and categories were blocked for in-district students who were using the internet; we did this by requesting additional information from districts we highlighted in the story and from the students who attend their schools.

When we calculated the total number of blocks represented in our data, we tallied all of them, regardless of whether we knew students could access websites in given categories. This project, after all, is about filter-enabled censorship, and we aimed to catalog the true scale of blocking, whomever it impacts.