No, Harvard Didn’t Correct Education Secretary’s Letter
PLUS: Meta AI’s Hallucination About Conservative Activist Leads to Defamation Suit; Kremlin Celebrates WWII Victory Day with Another Ukraine Hoax
Welcome to Reality Check, original reporting that’s your BS detector.
Our work is more important than ever. Please support us by becoming a Premium Member. And forward this email to a friend to spread the word!
Follow us on your social media platform of choice: X | LinkedIn | Instagram | Bluesky
In today’s edition, liberals praise Harvard after an MIT student doctored and then posted a letter to Harvard from the Trump administration to make it look like Harvard had redlined the letter with multiple edits correcting its grammar. And in one of the first suits of its kind a conservative activist sues a popular AI chatbot for defamation, while Voice of America, long rated as highly credible by NewsGuard, is adding reporting from One America News, which NewsGuard has found repeatedly posts false or egregiously misleading claims on subjects ranging from elections to FEMA to COVID.
Today’s newsletter was edited by Sofia Rubinson and Eric Effron.
1. Viral Posts Claim Harvard Responded to Threatening Letter from Education Department with a Tough Edit
By Sarah Komar

What happened: Liberal social media users are circulating a doctored version of a letter that Education Secretary Linda McMahon sent to Harvard University, falsely claiming that Harvard officials marked up the letter to correct McMahon’s congenitally shaky grammar.
Context: On May 5, McMahon posted a letter addressed to Harvard President Alan Garber, stating that the university “should no longer seek GRANTS from the federal government, since none will be provided.”
In the letter, McMahon accused Harvard of a range of alleged offenses, including illegal race-based affirmative action practices and admitting foreign students “who engage in violent behavior and show contempt” for the United States.
A closer look: The day after McMahon posted the letter to her X account, a version surfaced online, covered in red-pen edits pointing out incomplete clauses, inconsistent capitalization, and grammar slip-ups. Left-leaning accounts eagerly shared the spoof, praising Harvard for supposedly putting the secretary of education in her place. At the same time, some conservative users criticized Harvard for arrogantly treating McMahon as a lowly student.

Liberal X user @PrincessBravato posted the marked-up letter, stating: “I can’t stop laughing. Harvard sent back the Linda Letter marked for spelling & punctuation…. Morons, literal stupid people run the government. Very low IQ cabinet.” The post received 2.4 million views and 81,000 likes in a day.
Liberal Threads user @meidas_societyschild shared the letter and wrote: “Linda McMahon wrote Harvard a letter to which they responded by noting areas that are in need of corrections and then posted it on social media. Great job Harvard!” The post garnered 693,000 views and 8,700 likes in a day.
In response to liberals sharing the marked-up letter, conservative X user @HRH_SHP stated, “It just makes Harvard look even more condescending and elitist.” X user @False_Shadows wrote, “Mostly ridiculous markups and debatable usage — same people who couldn’t identify plagiarism,” apparently referring to a plagiarism scandal involving Harvard’s former President Claudine Gay.
Actually: The annotated letter wasn’t Harvard’s handiwork. It was created by X user Daniel Luo, a Ph.D. student at the Massachusetts Institute of Technology who has no connection to Harvard.
In an email to NewsGuard, Luo said that he “felt very perturbed by the writing style” of the education secretary’s letter and “thought it would be funny to mark-up the paper in the style of an exasperated high school teacher trying to give feedback to a struggling student.”
While Harvard did not edit McMahon’s letter, it did respond in a May 6 statement that accused the federal government of trying to “impose unprecedented and improper control over Harvard University.”
Become a Reality Check Member and Get These Exclusive Benefits
Want to get smarter about misinformation? Become a Reality Check Member today and you'll get:
A free copy of the definitive book on the misinformation crisis, The Death of Truth by bestselling author Steven Brill ($30 value)
Exclusive members-only content and digital briefings
Free access to NewsGuard's browser extension that shows reliability ratings for 10K+ news sites right in your browser ($25 value)
2. Conservative Activist Sues Meta AI Alleging False Claims About Him, Potentially Creating New Liability for AI Chatbots

At a time when artificial intelligence seems to be stepping into many roles previously held only by humans, perhaps it should not be shocking that an AI bot has become a defendant in a defamation suit.
What happened: Conservative activist and anti-DEI crusader Robby Starbuck filed a defamation suit against Meta, claiming that the company’s AI chatbot falsely accused him of being present at the Jan. 6, 2021, attack at the U.S. Capitol.
A closer look: NewsGuard traced the controversy to an August 2024 dustup between Starbuck and a Vermont motorcycle dealership called Wilkins Harley-Davidson. Starbuck criticized Harley-Davidson for the company’s DEI initiatives and corporate advocacy.
In the midst of the feud, the dealership posted a screenshot showing a Meta AI response (the prompt was not provided) that claimed, “Starbuck was present at the [U.S. Capitol] riot and has been accused of participating in or promoting the event.” The post included the caption, “Interesting what Meta AI says about you Mr. Conspiracy Theorist,” referring to Starbuck.
Actually: “None of this ever happened,” Starbuck stated in a video he posted on YouTube. “It is literally an invention of Meta’s AI.” Indeed, he has plenty of proof to back his denial.
On Jan. 6, 2021, Starbuck, who has lived in Franklin, Tennessee, since 2019, filmed a live video for his podcast “Grow Up” with his wife Landon Starbuck from their Tennessee studio.
That same day, Starbuck posted on X more than 40 times, calling Jan. 6, 2021, “one of the worst days in modern American history” and “shameful.” While the riot was ongoing, Starbuck posted, “I don’t care what group you’re a part of, destroying property is stupid & wrong.”
Starbuck does not appear in NPR’s searchable database of the more than 1,500 federal criminal cases related to the Jan. 6, 2021, riot.
See you in court: On April 28, 2025, Starbuck filed a defamation suit in Delaware Superior Court against Meta for “repeatedly publish[ing] — and continu[ing] to publish — provably false and defamatory statements,” according to a press release from Dhillon Law Group, the firm that is representing Starbuck. Its clients are often conservatives making civil rights claims, and its founder, Harmeet Dhillon, recently became assistant attorney general in charge of the Civil Rights Division in the Trump administration Justice Department.
Starbuck is seeking $5 million in damages.
On April 29, Meta’s chief global affairs officer Joel Kaplan issued an apology to Starbuck on X, stating: “This is unacceptable. This is clearly not how our AI should operate.”
Meta’s chatbot has since corrected the record. When NewsGuard asked Meta AI on May 7 whether Starbuck participated in the Jan. 6 insurrection, it responded, “Robby Starbuck was not at the January 6, 2021, Capitol riot.” The chatbot added, “Meta’s AI chatbot falsely claimed he participated in the riot, was arrested, and charged with a misdemeanor, prompting Starbuck to file a $5 million lawsuit against the company for spreading false information about him.”
Starbuck joins a small group of plaintiffs who have turned to the courts to try to hold AI companies accountable for generating allegedly false information. While none of the courts has yet awarded damages for defamation, The Wall Street Journal reported, some suits are advancing.
In one case, a Georgia judge last year allowed a defamation case against OpenAI to continue after conservative radio host Mark Walters claimed that its ChatGPT product inaccurately tied him to an embezzlement suit. OpenAI responded in April 2025 that it warns users that its chatbot can produce inaccuracies and said Walters did not alert the company about the alleged error before initiating the suit.
Click here to find out more about NewsGuard Trust Scores and our process for rating websites. You can download NewsGuard’s browser extension, which displays NewsGuard Trust Score icons next to links on search engines, social media feeds, and other platforms by clicking here.
3. Kremlin Sources Concoct WWII Victory Day Falsehood to Smear Ukraine
By Eva Maitland

What happened: Pro-Kremlin sources are spreading a fabricated image of a leaflet purportedly created by the Ukrainian government that advises World War II veterans to hide Soviet-era medals from public view, attempting to portray Ukraine as disrespectful to war vets.
Context: The leaflet was shared in the days leading up to Victory in Europe Day (May 8), which in Russia and some post-Soviet countries is known as Victory Day and celebrated on May 9. The occasion marks the defeat of Nazi Germany in World War II. Since 2023, Ukraine has celebrated the anniversary on May 8, in line with other European countries.
A closer look: The image shows what appears to be a leaflet, which stated in Ukrainian: “The Law of Ukraine ‘On the Condemnation of Communist and National Socialist (Nazi) Totalitarian Regimes in Ukraine and the Prohibition of Propaganda of Their Symbols’ prohibits the open wearing of Soviet medals and orders. If you are going to wear a Soviet award, attach it to the inside of your jacket, as shown in the diagram.”
The purported leaflet includes a diagram demonstrating how to wear the medals, along with the logo of Ukraine’s Ministry of National Memory, a government agency focused on policies related to Ukrainian history.
The earliest instance of the image shared online identified by NewsGuard was a May 5 Telegram post by @zovcrimea, a pro-Kremlin account that regularly posts Russian disinformation. “‘The Institute of National Memory in Ukraine’ mocks Ukrainian elderly,” the post stated, garnering 16,000 views in two days.
The image spread widely on social media and was amplified by sites in the Pravda network, a Kremlin-affiliated network of more than a hundred websites publishing Russian disinformation with an apparent aim to infect AI chatbots with false information, as previously reported in Reality Check.
Actually: The leaflet is not authentic, and Ukrainian war veterans are able to openly wear Soviet-era medals.
A spokesperson from the Ukrainian Institute of National Memory told NewsGuard in a May 2025 email that the agency “has no connection to this leaflet” and said that it is likely “part of Russian propaganda.” A NewsGuard review of the Institute’s website and official social media channels did not turn up any instructions for how to wear Soviet medals.
While Ukraine does have a law condemning Nazi and communist symbols, a NewsGuard review of the full text found that it explicitly states that the ban “does not apply” to “state awards, jubilee medals and other distinctions awarded to individuals before 1991 and … in connection with the anniversaries of World War II events, as well as on documents certifying the awarding of them.”
4. Voice of America (NewsGuard Trust Score: 100/100) Taps One America News (22.5/100) to Provide News Services
By Sam Howard and Sofia Rubinson
What happened: One America News (NewsGuard Trust Score: 22.5/100), a right-wing cable news network that NewsGuard has found to repeatedly air false and egregiously misleading information, has been tapped by U.S.-funded Voice of America (Trust Score: 100/100) to provide news programming.
A closer look: Kari Lake, senior adviser for the U.S. Agency for Global Media, which oversees VOA, announced the partnership in an X post on May 6, stating that the change will help ensure that Voice of America has “reliable and credible options as they work to craft their reporting and news programs.”
NewsGuard reviews of OAN’s website and TV programs dating to 2020 have found that the network frequently spreads false claims, particularly about U.S. politics and elections.
In the wake of the 2020 election, the OAN network and site became prominent proponents of baseless claims advanced by Donald Trump that he was the rightful winner of the election. As recently as November 2024, OAN reporter Pearson Sharp said on air: “The 2020 election was bad enough — we all know it was stolen. And friends, I’m afraid 2024 ain’t lookin’ any better. After all, if they figured out how to successfully steal an election last time, why on earth wouldn’t they do it again?”
NewsGuard also found OAN often publishes false claims related to health. For example, an April 2025 article published on the OAN site stated that Pfizer’s COVID-19 vaccine caused 23 percent more deaths in vaccinated people compared to a group that received a placebo in the vaccine’s clinical trial. The clinical trial results actually said that no deaths among study participants were caused by the vaccine.
Voice of America has received the highest NewsGuard trust score since it was first reviewed in 2020.
Articles on the VOA site have typically been fact-based and well-sourced, drawing on firsthand reporting, interviews with academics and other experts, government reports and data, and coverage from other reliable media organizations.
VOA, which is funded by Congress and broadcasts U.S. political news and stories about everyday life in America globally, has undergone significant changes during the second Trump administration.
In December 2024, Trump named Lake, a former TV anchor and Trump ally, to oversee U.S. government-funded media operations, which include VOA, Radio Free Europe, and Radio Martí, among others. Lake has previously promoted baseless election fraud claims and called for the punishment of “lying” journalists.
In March, Trump signed an executive order that placed nearly all of VOA’s 1,300 employees on administrative leave, although a federal judge has temporarily halted that move. Before tapping OAN, VOA ended its contracts with credible news agencies including The Associated Press (Trust Score: 100/100), Reuters (Trust Score: 100/100), and Agence France-Presse (Trust Score: 100/100).
Lake said that OAN will be providing its newsfeed to VOA free of charge.
In a May 2025 emailed statement to NewsGuard, OAN President Charles Herring stated, “Unlike NewsGuard, which took a reported $750,000 from the U.S. Govt, OAN has consistently refused any government funds to avoid so much as the appearance of a conflict of interest, diminishing its independence and journalist integrity. The recent announcement is no exception.” (NewsGuard had a $750,000 contract with the U.S. Defense Department in 2021 to aid in monitoring disinformation campaigns mounted against the United States by China, Russia, and Iran.)
NewsGuard sent an email to Lake’s office, seeking comment on the arrangement, but did not get an immediate response. NewsGuard will continue to monitor and analyze OAN and VOA based on NewsGuard’s nine apolitical credibility and transparency standards.
Reality Check members can read NewsGuard’s detailed Nutrition Labels for OAN and VOA here.
If you see something, say something
If you see or hear something that you think may be provably false, please alert NewsGuard via realitycheck@newsguardtech.com and we'll do our best to get to the bottom of it. Note: Tips should not include content that you simply disagree with, however strongly.
Reality Check is produced by Co-CEOs Steven Brill and Gordon Crovitz, and the NewsGuard team.
We launched Reality Check after seeing how much interest there is in our work beyond the business and tech communities that we serve. Subscribe to this newsletter to support our apolitical mission to counter misinformation for readers, brands, and democracies. Our work is more important than ever. Have feedback? Send us an email: realitycheck@newsguardtech.com.