Is it true that the photo of a burned Israeli child was created by a neural network based on a photo of a rescued dog?

On October 13, 2023, statements appeared on social networks that the photograph of a child killed at the hands of Hamas militants, shown by the Israeli Prime Minister, was a fake. It is reported that Internet users allegedly found the original, which depicts a living puppy, and not a charred body. We have verified the accuracy of these reports.

October 12, 2023 on the official English-language account of the Prime Minister of Israel on the social network X (formerly Twitter) published three photos with the following caption: “Here are some photos that Prime Minister Benjamin Netanyahu showed US Secretary of State Antony Blinken. Warning: These are horrifying photographs of babies being killed and burned by Hamas monsters." The next day, publications circulated on the Internet claiming that at least one of these photographs was a fake.

For example, in the Telegram channel of Channel One host Ruslan Ostashko (115,000 views at the time of writing this analysis) stated: “But there is a nuance. The original photo, which was found on Reddit, shows a dog. And artificial intelligence turned it into pseudo-proof.” Channel Colonelcassad (217,000 views) also commented allegedly fabricated photo: “Such a number of fakes from the Israeli authorities and the Israeli media indicates a serious lack of original materials, since such fabrications have to be dealt with.” TV channel "Tsargrad" added: “As you know, the first victim of war is always the truth.” The authors of many Russian-language publications called the story of exposing the alleged forgery a “failure of Israeli propaganda” - similar posts appeared in Telegram channels “KB +18"(225,000 views), Voblya 18+ (168,000), "Belarusian security officer"(136,000), "Oleg Tsarev"(66,000), "Gossip"(42,000), "Ramsay" (37,000), "Ax | Policy" (31,000), "DPR Info" (27,000) and "Therapy"(26,000), as well as on VKontakte (examples here, here, here, here And here).

From the editor: We must warn readers that this material shows a photo published by the Israeli authorities. This is a truly scary image. If you don't want to see it, please don't read the analysis.

Photo: screenshot post from the TGStat service

The radical Islamist movement Hamas (in Israel, the USA, the EU, Japan and a number of other countries it recognized terrorist organization) on the morning of October 7, 2023 caused massive rocket attacks on Israel from the Gaza Strip. At the same time, militants infiltrated nearby Israeli settlements, where they attacked civilians, and military installations. Soon appeared reports that among the dead there are women, elderly people and minors - in particular, the representative of the Israel Defense Forces, Jonathan Conricus, citing data from a forensic expert statedwhat's in kibbutz Beeri, the bodies of headless babies were found. However, Hamas previously denied that militants killed children during the attack. As of October 13, the attack killed more than 1300 Israelis, and Israel's response - more 1,500 Palestinians.

On October 12, shortly after the publication appeared on the account of the Israeli Prime Minister, one of the photographs reposted on the social network X (formerly Twitter) American conservative political commentator Ben Shapiro. “Did you need visual evidence of dead Jewish babies? That's it, pathetic Jew haters. Israel minimizes civilian casualties. But Israel will not allow the pieces of human shit that did this to live. Every ounce of blood shed in Gaza belongs to Hamas,” he wrote. The recording received more than 14 million views. At the same time, some users in replays doubted in the authenticity of the photograph due to the strange-looking brush of the person showing the body of the child. It's all about the rubber glove, which was not fully stretched on the right hand, which is why the fingers of the captured person seem deformed. But some commentators decidedthat the picture was created by a neural network, since strange-looking limbs are one of the main visual indicators AI art.

Photo: screenshots of tweets

The next day, Jackson Hinkle (American political blogger adhering to pro-Russian positions) also accused Shapiro in publishing an allegedly fake photograph. “The image that Ben Shapiro tried to pass off as the burned corpse of a child was a fake image created by AI!” - he wrote on the social network X, where the post received almost 8 million views. Hinkle supported his position with a screenshot with the results of photo analysis in the service AI or Not, which, according to the creators, determines with high accuracy whether an image was created by AI or a human. “Verified” also uploaded the photo to AI or Not, and according to the analysis, the photo of the dead child was allegedly generated by a neural network. Under Shapiro's tweet even appeared Community Notes - a note about context, which, according to some readers, will be useful to others) with the clarification that the photo was fabricated. However, this plaque was later removed.

The fact is that in reality the work of the AI ​​or Not service is far from perfect and it can hardly be called highly accurate. In replays to Shapiro and Hinkle’s tweets, many users shared their screenshots with the results of analysis of other photos of themselves: in some cases service installedthat the real workings of neural networks were created by man, and in others — that real photographs were created using AI. Moreover, some users, when analyzing the same photo of a burned baby, received from AI or Not another result: the photo was not created by AI. These findings, at the time of writing, are reflected in the updated Community Notes for Hinkle’s tweet. Back in the summer of 2023, in an article by The New York Times compared the work of several services for identifying AI images and found that some detectors (including the one used by Hinkle and other bloggers) do not always give the correct results. Representatives for AI or Not did not respond to journalists’ requests for comment at the time.

Photo: screenshot tweet

As for the photo with the dog, it went viral after many users steel send her in comments to Hinkle and Shapiro's tweets, calling The original is the photograph with the animal. As in most of the publications mentioned at the beginning of the analysis, the photo with the puppy is not an independent image, but a screenshot of a post on a certain website. Despite numerous claims that the supposed original was discovered by Reddit users, the screenshot itself shows the anonymous web forum 4chan. This is clear not only from the design and interface - if you google the phrase from the post, then extradition Google will be on 4chan, where the photo with the animal was posted on thread entitled “Ben Shapiro uses artificial intelligence-generated image to discredit HOMOS” (possibly an untranslatable pun in the original: HAMAS - Hamas and Homos - “homosexuals”). Anonymous user wrote: “Here's the original photo, they used a puppy rescue photo as a base and the AI ​​did the rest. Source: dad works at Centcom (Central Command of the US Armed Forces. - Ed.).”

The Arlington Post Twitter account conducted an error rate analysis (E.L.A., a popular method for assessing the authenticity of images) in a photo of a dog and a photo of a body published by the Israeli Prime Minister. Online edition concludedthat the photograph of the deceased child had not been altered, as evidenced by consistent and uniform compression artifacts, sharp edges, noise patterns and brightness levels. But the photo with the dog shows clear signs of digital processing, especially in suspicious areas along the contour of the dog: the edges look sharper and brighter, which indicates probable manipulation.

Photo: screenshot tweet

“Verified” found out that the original source was not a post on 4chan, but a tweet from a user under the nickname Stellar Man, accompanied by a photograph and published after Netanyahu’s post, wrote (the message has already been deleted, but remains copy, saved in cache Google): “Took me five seconds to do this. Creating a convincing fake photo is no longer difficult. It’s harder to fake a video, but they’ll never show a video with proof.” When a photo of a dog went viral, Stellar Man noticed: “What have I done, lmao (laughing my ass off, “laughing until I drop.” - Ed.).” As verified, the picture was posted on 4chan approximately 45 minutes after the tweet.

Photo: collage “Verified” / social networks

Thus, the photo with the dead child, which is presented as the work of a neural network, is in fact genuine, and the photo with the dog, called the altered original, was created using artificial intelligence.

Cover photo: social networks / @stellarman22

Read on the topic:

  1. Is it true that the video shows Israeli children abducted by Hamas on October 7?
  2. Is it true that there are such posters with Putin in the Gaza Strip?
  3. Is it true that the video shows the consequences of the shelling of Gaza?
  4. Putin on his knee, the Pope's down jacket and Trump's arrest: how to recognize viral photos created by neural networks
  5. Is it true that a Polish TV channel used a neural network to erase Harry Potter’s scar because of its resemblance to the letter Z?

If you find a spelling or grammatical error, please let us know by highlighting the error text and clicking Ctrl+Enter.

Share with friends

Typo message

Our editors will receive the following text: