AI Image of 24-Year-Old 9/11 Hero Welles Crowther Goes Viral, Distorting History

Q: As the 25th anniversary of 9/11 arrives, can we still trust our own eyes when AI rewrites historical visual evidence?

No. As AI gets better at fabricating plausible “historical” images, eyesight alone is no longer a reliable test of authenticity. The Welles Crowther case shows why. Crowther’s heroism on 9/11 is well established through survivor testimony and later honors, but there is no known authentic public photo of him rescuing people inside the South Tower. The viral image claiming to show that moment appears to be AI-generated. Yet several AI tools gave misleading answers, with some treating it as a real photo or a reenactment, and Google’s AI overview even misidentified it as a scene from an unrelated miniseries. That means the same systems people may use to verify images can also deepen confusion. The deeper risk is to historical memory. Nearly a quarter century after 9/11, many Americans have no living memory of the attacks and encounter them mainly through online images, clips and summaries. A fabricated picture can attach itself to a real story and gradually become accepted as visual evidence, especially when no genuine image exists to disprove it. The safeguard is not to reject visuals entirely, but to treat them as claims that require corroboration. For 9/11, the strongest record remains original reporting, authenticated photographs and video, official investigations such as the 9/11 Commission and NIST reports, museum archives, and firsthand witness accounts. In this environment, trust should shift from what looks real to what can be sourced, dated, and independently verified.

1 articles · Updated · Gizmodo · Sep 11

A fabricated image claiming to show Welles Crowther rescuing people in the World Trade Center is spreading online, despite no known authentic photos of the 24-year-old in action on Sept. 11.
Crowther’s heroism is well documented by witnesses—he is credited with helping save at least a dozen people from the South Tower before dying in its collapse—but the viral picture is not genuine.
Major AI tools muddied verification: ChatGPT, Grok and Claude initially treated the image as real or a dramatization, while Gemini first declined to check for Google’s SynthID watermark and then found none.
The false image highlights how cheap, fast AI generation can reshape public memory of 9/11, especially with roughly 25% of Americans alive in 2026 born after the 2001 attacks.