A video message from the President of the United States about the failure of the lunar mission in 1969 has been published. It shows how deepfakes work
The Apollo 11 moon landing on July 20, 1969, was a landmark moment in space history. But what if the astronauts had died on the flight to the moon, and US President Richard Nixon had to bring this tragic news to the Americans on television?
In a video posted on a special website, which looks frighteningly convincing, President Nixon allegedly reveals that NASA failed and the astronauts died on the Moon. Deepfake (lit. – “Deep fake”) – is falsified videos in which people using AI do things that they never did. Sometimes such fakes are difficult to distinguish from real videos.
“Fate dictated that people who went to the moon to explore the world would stay on the moon to rest in peace),” said Mr. Nixon in a fake video about astronauts Neil Armstrong, Buzz Aldrin, and Michael Collins (Michael Collins).
AI experts at MIT took six months to create a very convincing 7-minute fake video, in which real NASA footage is interspersed with a fake Nixon tragic speech about the failure of the Apollo 11 mission.
Deep-learning artificial intelligence technology was used to make Nixon’s voice and facial movements compelling. By the way, the voiced tragic speech is real – it was prepared in case of the death of astronauts and is stored in the US National Archives.
The Massachusetts Institute of Technology created the Event of Moon Disaster to show people the dangerous impact fake videos can have on an unsuspecting public. “By creating this alternate history, the project explores the impact and prevalence of misinformation and false technology in our modern society,” – noted in the project site.
In the case of the Event of Moon Disaster, the goal is not only to help people better understand the Deepfake phenomenon, but also to explain how fakes are made, how they work, how to identify them; assess their potential use and abuse, and develop means to combat counterfeiting and misinformation. This project was supported by a grant from the Mozilla Creative Media Awards.
