No, That U.S. Government Go-Kart Image Does Not Show Trump With Hillary Clinton and Jeffrey Epstein
A bizarre AI-generated image posted by the U.S. Department of Education was widely interpreted as showing Donald Trump riding with Hillary Clinton and Jeffrey Epstein. The agency says the intended figures were Education Secretary Linda McMahon and Transportation Secretary Sean Duffy—and the image was generated with AI.
An image shared by an official U.S. government account has created one of the strangest political-media moments of the week.
It appears, at first glance, to show Donald Trump riding in Mario Kart-style racing vehicles with a woman who looks like Hillary Clinton and a grey-haired man who looks remarkably like Jeffrey Epstein.
That is not what the image was intended to show.
The Department of Education says the woman is Education Secretary Linda McMahon and the man is Transportation Secretary Sean Duffy.
The image was generated using artificial intelligence as promotional material connected to Washington’s Freedom 250 Grand Prix celebrations.
The confusion is not simply internet trolling.
The AI likenesses are poor enough that prominent political accounts and ordinary users independently made the Clinton and Epstein comparison.
That transformed an upbeat government promotional image into an accidental Epstein controversy.
The Daily Beast reports the graphic was generated using Grok Imagine and posted by the Education Department.
Duffy is deeply involved with the Freedom 250 event, a real IndyCar race taking place around Washington as part of America’s 250th anniversary celebrations.
McMahon’s appearance in the image was apparently meant to connect the Education Department to the broader patriotic event.
Instead, AI facial ambiguity took over the story.
This is a useful example of why synthetic government communications create a different level of risk from memes posted by anonymous users.
When an official federal account publishes an image, audiences naturally assume someone reviewed it.
If the AI turns a cabinet official into a person resembling one of the most notorious sex offenders in modern American history, the mistake carries political meaning whether the resemblance was intended or not.
The Trump-Epstein connection makes that sensitivity much stronger.
Trump and Epstein were socially acquainted decades ago, a relationship documented in authentic photographs and public reporting.
Trump later said they fell out and has denied wrongdoing connected to Epstein.
The political controversy over Epstein files and disclosure has continued for years.
That history means any image that appears to place Trump and Epstein together will spread immediately.
The same is true for Hillary Clinton, whose family has also been dragged repeatedly into real and fabricated Epstein imagery online.
AI-generated political pictures thrive precisely because viewers already possess these associations.
A synthetic image does not need to be perfect.
It only needs to resemble the story people already expect.
There is also an important distinction between this image and a deliberate deepfake.
There is no evidence the Education Department intentionally tried to depict Epstein or Clinton.
The problem appears to be bad generative output and inadequate review.
That makes the episode less sinister and arguably more revealing.
Government agencies are beginning to use consumer generative-AI tools for public communications, but those systems do not understand reputational context the way a human communications team should.
A model can produce a face statistically similar to the prompt while accidentally making it look like another famous person.
A human reviewer needs to catch that before publication.
The incident also shows why reverse-image and provenance checks are becoming essential.
Someone seeing a screenshot stripped of the original post could easily believe it was evidence of a real event.
The Freedom 250 race is real.
Trump is genuinely participating ceremonially.
Sean Duffy is genuinely involved.
The go-kart-style scene is synthetic.
No authentic event placed Trump, Hillary Clinton and Jeffrey Epstein together in those vehicles.
This is the new misinformation problem: a government can publish something synthetic for harmless promotional reasons, social media can misidentify the synthetic people, and within minutes the image becomes ‘proof’ of an event that never happened.
The open question is whether agencies learn from this embarrassment and introduce stronger human review and AI-labeling rules—or whether official accounts themselves become another source of viral synthetic imagery that citizens have to fact-check before believing their own eyes.