AI-generated images misrepresent details of Toronto airport accident

Shocking footage of a plane flipping over on the runway at Toronto Pearson International Airport spread online as news of the accident broke, but one image supposedly showing emergency workers putting out fires and another allegedly depicting the craft's pilot are not legitimate. Watermarks indicate both images were generated by artificial intelligence and an AI detection tool also found them likely to be digital creations.

"US flight 4819 of Delta Airlines crash lands UPSIDE DOWN in Toronto," reads the caption of a February 17, 2025 X post sharing an image that appears to show an inverted airplane with first responders extinguishing a nearby fire.

The same image spread across X, Instagram, YouTube and in a handful of articles. The picture also accompanied online discussions on the accident in Arabic, Greek and Spanish.

Image
Screenshot of an X post taken February 19, 2025

Delta Air Lines said all 76 passengers and four crew members survived the crash-landing which flipped a plane upside-down on February 17 at Toronto Pearson. The cause of the accident is currently under investigation, but multiple videos reportedly taken by people at the scene show the Bombardier CRJ-900 aircraft skidding on the tarmac, turning over and appearing to catch on fire. 

However, the image of workers supposedly putting out flames near the inverted aircraft is not authentic. This version of the image bears a watermark for GrokAI, indicating it was generated using the artificial intelligence chatbot available on X.

The supposed yellow emergency vehicle seen in the image displays no legible writing or insignia and does not appear in the multiple videos and images taken by AFP at the scene of the accident.

Image
Screenshot of an image in an X post taken February 19, 2025 with highlights added by AFP
Image
Red fire trucks respond to the Delta airlines plane that crash landed at Toronto Pearson Airport in Toronto, Ontario, on February 17, 2025 (Geoff Robins)

AFP also used the AI detection tool Hive, which found there was a 98.7 percent chance the image was generated using artificial intelligence.

Image
Screenshot from Hive taken February 19, 2025

AFP previously fact-checked AI-generated images of the January 29 air crash near Ronald Reagan Washington National Airport.

Stolen satire pilot photo

A second image was widely shared with claims it showed "Jonathan Simpson" the pilot of the plane.

In an Instagram reel, a user claims a man seen in a uniform was identified as the pilot of the aircraft when it landed at Pearson Airport. The same image, accompanied by his supposed name, spread on X and Facebook, also appearing in Spanish, German and Portuguese. One post included the image of the alleged pilot and the AI-generated tarmac response shot in a collage alongside verified images from survivors at the scene.

Image
Screenshot of an Instagram post taken February 18, 2025

However, a reverse image search reveals an earlier publication with the picture, also watermarked with a GrokAI logo. The logo on the pilot's uniform does not match Delta's insignia (archived here).

The AI detection tool Hive estimated there was a 99.5 percent chance the pilot's image was created with artificial intelligence.

Image
Screenshot generated from Hive on February 19, 2025

The account which originally published the image is marked as parody and frequently shares humorous posts about "Jonathan Simpson" in notable situations while using images of an adult film star. While some users posting the claim appeared to be in on the joke, others seemed to credulously share the picture without an explanation that it was satire.

Image
Screenshot of an X post taken February 19, 2025 with highlights added by AFP

The name of the pilot had not been made public as of 2000 GMT on February 19 (archived here).

Similarly, AFP has previously fact-checked claims that misinterpret a long-running hoax placing comedian Sam Hyde at the center of current events. 

Read more of AFP's reporting on misinformation in Canada here.

Is there content that you would like AFP to fact-check? Get in touch.

Contact us