“Nasa would have lied about the photos and videos of the moon? “. This old conspiracy theory, which claims that the Americans staged the images of astronauts on the moon, resurfaced recently on social networks. The cause ? A video showing Russian President Vladimir Putin being presented with the results of an image verification artificial intelligence that allegedly concluded that photos taken on the moon were fake.
For several weeks, numerous publications have taken up this assertion, such as this video viewed more than 274,000 times on TikTok, or this onepublished on the same social network, which states that “Man has never walked on the moon according to Google’s AI”.
The latter states in particular: “It turns out that the AI itself has serious doubts about the Apollo 11 mission. Russian President Vladimir Putin himself was able to view the analysis carried out by Google’s AI. If we accept the result, the images showing the moon landing of American astronauts may not be authentic. »
Videos of the same type are also shared in English or even in Spanishusing the same elements, also on other social networks like X. Some of these publications explain that the Russian president would have been “invited to Google”, and would have himself asked the engineers of the American company to test “the video of the landing on the moon” with their artificial intelligence for verification.
These different publications, which repeat the same narrative with sometimes some variations, also often claim that a few days later, NASA published a press release to justify itself, explaining that they would have lost the original images of the first steps on the moon. Recently, the entrepreneur Oussama Amar, very popular on social networks, also relayed this information in one of his podcasts ” Without permission “.
FAKE OFF
Video from Putin’s visit to AI conference
The video dates from November 24, 2023, during a visit by the Russian president to a conference on the theme of artificial intelligence organized in Moscow entitled: “Journey into the world of artificial intelligence”, as reported by the agency Press Tass.
Traces of the video can be found in the publications of the Russian press agency RIA Novosti, which uses it as the basis for an article titled: “Putin briefed on neural network findings on US moon landing” on November 24. The sequence was published on the agency’s Telegram channel the same day at 1:55 p.m.
What does the video really show?
As the article indicates, in these images Putin is accompanied by German Oskarovitch Gref, chairman of the board of directors of Sberbank, the Russian public bank which is organizing the event. The video, 27 seconds long, does not allow you to see the start of the sequence, with the images of the astronauts in question.
A more complete sequence can be found in a video from the Russian news agency Sputnik published on the Chinese social network Weibo. It is Nikolai Gerasimenko, presented by RIA Novost as “head of data research at Sberbank”, who first addresses Putin.
“Almost all objects in this image are considered fake [par l’intelligence artificielle] » , he says first, talking about the photos of the astronauts. “On the other hand, this photo was taken by the Chinese lunar rover. It does not pose any particular problem,” he shows afterwards. “The image is considered fake? »Putin intervenes. “Yes, look at the red.” “That’s what Google’s neural network thinks, not ours. There is therefore no bias. », Specifies German Oskarovitch Gref. “Very interesting,” Putin comments shortly after.
This video was therefore not filmed “at Google”, but rather during a Russian conference. Only the tool discussed in the sequence is referred to as a Google “neural network”. The people present around Putin are therefore not employees of the American tech giant either.
The photos are real shots from NASA
Numerous online publications explain that this sequence would call into question the veracity of the images of the Apollo 11 mission, which marked man’s first steps on the moon in 1969. However, the photos used during the presentation to the Russian president are those of the Apollo 15 mission in 1971, the fifth mission to send American astronauts to the surface of the moon.
These photos are shots taken during this mission, and are available on the NASA website. There left photo shows astronaut James Irwin digging a trench with a shovel, as explained in the photo description on the NASA website. There photo on the right shows astronaut David Scott using a drill.
On the other hand, NASA has not published any press release on this subject to try to justify itself following the broadcast of the sequence, as the agency confirmed with AFP. An article from the Reuters news agency, often used as proof of an admission by NASA, actually dates from 2009 and deals with images from the Apollo 11 expedition, which is therefore not the one from which these photos came.
Caution on the reliability of image verification tools
It is possible to use several tools to identify the origin or veracity of an image, such as 20 minutes has already presented it in different articles. More recently, given the development of image generation tools using artificial intelligence, there are also techniques and automated tools that can help judge the authenticity of a photo.
But, in general, it is advisable to remain cautious about the results given by tools using artificial intelligence to verify an image. The specialist in images generated by AI Victor Baait reminded the‘AFP that these tools were regularly wrong: “Verification tools can claim that an image is artificial intelligence when it is not. We must keep in mind that these tools are only indicators and we must not have blind confidence in them at all.”
The lack of information on this “neural network” is also a gap in judging its reliability, because we do not really know what parameters are analyzed by the tool. “If it is a real tool trained for the detection of fake news, NASA images being so widely used in fake news forums that a neural network which has been told that any image coming from these forums are fakes, would therefore classify them as such”, specifies with 20 minutes Emmanuel Adam, spokesperson for the French Association for Artificial Intelligence and lecturer at the Université Polytechnique Hauts-de-France specializing in AI.