Latest news with #HanyFarid
Yahoo
6 hours ago
- Yahoo
AI-generated videos fuel falsehoods about Iran-Israel conflict
In recent days, videos generated by artificial intelligence have surfaced online purporting to show dramatic scenes from the Iran-Israel conflict, including an AI-generated woman reporting from a burning prison in Tehran and fake footage of high-rise buildings reduced to rubble in Tel Aviv. Other fabricated visuals depict a downed Israeli military aircraft. These clips, some which have racked up millions of views on platforms including X and TikTok, are the latest in a growing pattern of AI-generated videos that spread during major events. Researchers at Clemson University's Media Forensics Hub told CBS News that some of the content is being amplified on X by a coordinated network of accounts promoting Iranian opposition messaging — with the goal of undermining confidence in the Iranian government. Fabricated videos On Monday, Israel carried out strikes on several sites in Iran, including the notorious Evin Prison. Within minutes of the attack, a video began circulating on X and other social media platforms showing an explosion at the entrance. The video is grainy, black-and-white and appears to be security camera footage. But several visual anomalies indicate the footage may have been created using artificial intelligence, experts say, including an incorrect sign above the door and inconsistencies with the explosion. Hany Farid, a professor at the University of California, Berkeley, and co-founder of AI detection startup GetReal Labs, told CBS News he believes the video may have been generated by an AI image-to-video tool. Farid said recent advancements in technology have helped lead to more realistic-looking videos with easier ways to create and share them quickly. "A year ago it was [that] you could make a single image that was pretty photo realistic," Farid said. "Now it's full blown video with explosions, with what looks like handheld mobile device imaging." The video had been posted on X within minutes of the June 23 Israeli attack on the facility by an account that "bears marks of being inauthentic," according to Media Forensics Hub researchers. Iranian and Israeli officials have not commented on the authenticity of the video. Darren Linvill, co-director of the Media Forensics Hub, told CBS News another video, which depicted an AI-generated reporter outside the prison, is the "perfect example" of a coordinated network using AI to circulate false information to wider audiences. "It isn't doing anything that one couldn't do with previous technology, it's just doing it all cheaper, faster, and at greater scale," Linvill said. It's not clear who is behind the videos, Linvill said. Social media platforms' responses When asked about the AI-generated Iran-Israel videos on their platform, a TikTok spokesperson told CBS News the platform does not allow harmful misinformation or AI-generated content of fakes authoritative sources or crisis events, and has removed some of these videos. A spokesperson for X referred CBS News to their Community Notes feature, and said some of the AI-generated video posts have had Community Notes added to help combat the false information. As for how to avoid falling prey to videos created with AI, Farid said, "Stop getting your news from social media, particularly on breaking events like this." Hegseth slams Iran strikes initial assessment that contradicts Trump's take Young Cuban girl asks Trump to lift travel ban stopping her from joining mom in U.S. Jeff Bezos and Lauren Sánchez set for star-studded wedding in Venice
CBS News
15 hours ago
- CBS News
AI-generated videos are fueling falsehoods about Iran-Israel conflict, researchers say
In recent days, videos generated by artificial intelligence have surfaced online purporting to show dramatic scenes from the Iran-Israel conflict, including an AI-generated woman reporting from a burning prison in Tehran and fake footage of high-rise buildings reduced to rubble in Tel Aviv. Other fabricated visuals depict a downed Israeli military aircraft. These clips, some which have racked up millions of views on platforms including X and TikTok, are the latest in a growing pattern of AI-generated videos that spread during major events. Screenshots taken by CBS News. CBS News Researchers at Clemson University's Media Forensics Hub told CBS News that some of the content is being amplified on X by a coordinated network of accounts promoting Iranian opposition messaging — with the goal of undermining confidence in the Iranian government. Fabricated videos On Monday, Israel carried out strikes on several sites in Iran, including the notorious Evin Prison. Within minutes of the attack, a video began circulating on X and other social media platforms showing an explosion at the entrance. The video is grainy, black-and-white and appears to be security camera footage. But several visual anomalies indicate the footage may have been created using artificial intelligence, experts say, including an incorrect sign above the door and inconsistencies with the explosion. Hany Farid, a professor at the University of California, Berkeley, and co-founder of AI detection startup GetReal Labs, told CBS News he believes the video may have been generated by an AI image-to-video tool. Farid said recent advancements in technology have helped lead to more realistic-looking videos with easier ways to create and share them quickly. "A year ago it was [that] you could make a single image that was pretty photo realistic," Farid said. "Now it's full blown video with explosions, with what looks like handheld mobile device imaging." The video had been posted on X within minutes of the June 23 Israeli attack on the facility by an account that "bears marks of being inauthentic," according to Media Forensics Hub researchers. Iranian and Israeli officials have not commented on the authenticity of the video. Darren Linvill, co-director of the Media Forensics Hub, told CBS News another video, which depicted an AI-generated reporter outside the prison, is the "perfect example" of a coordinated network using AI to circulate false information to wider audiences. "It isn't doing anything that one couldn't do with previous technology, it's just doing it all cheaper, faster, and at greater scale," Linvill said. It's not clear who is behind the videos, Linvill said. Social media platforms' responses When asked about the AI-generated Iran-Israel videos on their platform, a TikTok spokesperson told CBS News the platform does not allow harmful misinformation or AI-generated content of fakes authoritative sources or crisis events, and has removed some of these videos. A spokesperson for X referred CBS News to their Community Notes feature, and said some of the AI-generated video posts have had Community Notes added to help combat the false information. As for how to avoid falling prey to videos created with AI, Farid said, "Stop getting your news from social media, particularly on breaking events like this."
DW
21 hours ago
- Politics
- DW
Fact check: Viral Evin prison blast video is likely AI-fake – DW – 06/27/2025
A viral video claiming to show an Israeli strike on Iran's Evin prison is probably AI-generated. Experts reveal key visual inconsistencies. A video went viral this week, allegedly showing the footage of a surveillance camera filming the entrance to the notorious Evin prison in Tehran being destroyed by an explosion. According to Israel, it had fired on the prison; Iran's judiciary confirmed the attack. The prison is known for reports of human rights violations against political prisoners of the Iranian regime. Israeli Foreign Minister Gideon Saar had shared the black-and-white video (archived here) on his X account, writing: "We warned Iran time and again: stop targeting civilians! They continued, including this morning. Our response: Viva la libertad, carajo!" Numerous international media outlets featured or referenced the six-second video to report on the attack. These included the New York Times, BBC, and the German channel ARD. They have since deleted the footage and added disclaimers to their publications. But the video appears to be fake. Most probably generated with AI with the help of an old image of the prison gate. DW Fact check looked at the puzzle pieces. A photo of the prison entrance, which is identical down to the smallest details, was used in a Persian-language article by Voice of America from May 2023. It shows a larger section of the image and was presumably taken from this post on X, from January 2023. If you compare the video circulating now and the image from 2023 you will see that there are a lot of similarities which are just too exact for pictures of the same location that have been taken at least 2 years apart: Compare the missing tiles on the left side of the entrance, or the bare shrubs in the right lower corner of the marked in both images. The shrubs are also an indication that the photo cannot be recent, but was instead taken in winter, as there are no leaves at all. Other pictures of Evin Prison on the internet, which were actually taken in summer, show the bushes and shrubs covered in dense foliage. One example, presumably from 2021, can be found here. Also, images of the damage actually caused at the prison now, shared by Iranian news agencies, show trees and bushes covered in leaves, and not bare, as in the black and white video sequence. Hany Farid, a professor at the University of California, Berkeley, who specializes in digital forensics, also pointed out the inconsistencies in the video shared by Saar in a post on LinkedIn on Tuesday. He wrote: "Unfortunately, the video is heavily compressed, making forensic analysis challenging. (…) It seems more likely that an AI-powered image-to-video generator was used with this image as the source The poor video quality seems plausible by faking a surveillance camera recording and complicates reverse image searches, as does the change in color footage to black and white. Farid added: "If this video is, in fact, fake, then it adds to a growing and disturbing trend of fake content circulating online as major world events unfold, making our understanding of what is happening and how to respond shaky, at best." Various fact checks, for example by Libération, VRT and ABC News Australia, have already come to the same conclusion: the video is not authentic. However, many users online and media outlets fell for it because there really was an attack on the prison. According to Iranian media, the attack not only struck the prison's main entrance but also damaged a prosecutor's office and two courts within Evin, cited in media reports. Further confusion was caused because the viral video matches real footage of the entrance gate. Farid told DW in a previous interview, "The problem with the fake content is that it's not just that people are creating fake content. It's that they are muddying the waters and suddenly everything is suspect." With the improvement of AI-generating programs and the sheer amount of content shared online, this "muddying of waters" is making it more and more difficult to distinguish between what is real and fake.
The Star
4 days ago
- Business
- The Star
Tech-fueled misinformation distorts Iran-Israel fighting
WASHINGTON: AI deepfakes, video game footage passed off as real combat, and chatbot-generated falsehoods – such tech-enabled misinformation is distorting the Israel-Iran conflict, fueling a war of narratives across social media. The information warfare unfolding alongside ground combat – sparked by Israel's strikes on Iran's nuclear facilities and military leadership – underscores a digital crisis in the age of rapidly advancing AI tools that have blurred the lines between truth and fabrication. The surge in wartime misinformation has exposed an urgent need for stronger detection tools, experts say, as major tech platforms have largely weakened safeguards by scaling back content moderation and reducing reliance on human fact-checkers. After Iran struck Israel with barrages of missiles last week, AI-generated videos falsely claimed to show damage inflicted on Tel Aviv and Ben Gurion Airport. The videos were widely shared across Facebook, Instagram and X. Using a reverse image search, AFP's fact-checkers found that the clips were originally posted by a TikTok account that produces AI-generated content. There has been a "surge in generative AI misinformation, specifically related to the Iran-Israel conflict," Ken Jon Miyachi, founder of the Austin-based firm BitMindAI, told AFP. "These tools are being leveraged to manipulate public perception, often amplifying divisive or misleading narratives with unprecedented scale and sophistication." 'Photo-realism' GetReal Security, a US company focused on detecting manipulated media including AI deepfakes, also identified a wave of fabricated videos related to the Israel-Iran conflict. The company linked the visually compelling videos – depicting apocalyptic scenes of war-damaged Israeli aircraft and buildings as well as Iranian missiles mounted on a trailer – to Google's Veo 3 AI generator, known for hyper-realistic visuals. The Veo watermark is visible at the bottom of an online video posted by the news outlet Tehran Times, which claims to show "the moment an Iranian missile" struck Tel Aviv. "It is no surprise that as generative-AI tools continue to improve in photo-realism, they are being misused to spread misinformation and sow confusion," said Hany Farid, the co-founder of GetReal Security and a professor at the University of California, Berkeley. Farid offered one tip to spot such deepfakes: the Veo 3 videos were normally eight seconds in length or a combination of clips of a similar duration. "This eight-second limit obviously doesn't prove a video is fake, but should be a good reason to give you pause and fact-check before you re-share," he said. The falsehoods are not confined to social media. Disinformation watchdog NewsGuard has identified 51 websites that have advanced more than a dozen false claims – ranging from AI-generated photos purporting to show mass destruction in Tel Aviv to fabricated reports of Iran capturing Israeli pilots. Sources spreading these false narratives include Iranian military-linked Telegram channels and state media sources affiliated with the Islamic Republic of Iran Broadcasting (IRIB), sanctioned by the US Treasury Department, NewsGuard said. 'Control the narrative' "We're seeing a flood of false claims and ordinary Iranians appear to be the core targeted audience," McKenzie Sadeghi, a researcher with NewsGuard, told AFP. Sadeghi described Iranian citizens as "trapped in a sealed information environment," where state media outlets dominate in a chaotic attempt to "control the narrative." Iran itself claimed to be a victim of tech manipulation, with local media reporting that Israel briefly hacked a state television broadcast, airing footage of women's protests and urging people to take to the streets. Adding to the information chaos were online clips lifted from war-themed video games. AFP's fact-checkers identified one such clip posted on X, which falsely claimed to show an Israeli jet being shot down by Iran. The footage bore striking similarities to the military simulation game Arma 3. Israel's military has rejected Iranian media reports claiming its fighter jets were downed over Iran as "fake news." Chatbots such as xAI's Grok, which online users are increasingly turning to for instant fact-checking, falsely identified some of the manipulated visuals as real, researchers said. "This highlights a broader crisis in today's online information landscape: the erosion of trust in digital content," BitMindAI's Miyachi said. "There is an urgent need for better detection tools, media literacy, and platform accountability to safeguard the integrity of public discourse." – AFP
Yahoo
7 days ago
- Yahoo
Tech-fueled misinformation distorts Iran-Israel fighting
AI deepfakes, video game footage passed off as real combat, and chatbot-generated falsehoods -- such tech-enabled misinformation is distorting the Israel-Iran conflict, fueling a war of narratives across social media. The information warfare unfolding alongside ground combat -- sparked by Israel's strikes on Iran's nuclear facilities and military leadership -- underscores a digital crisis in the age of rapidly advancing AI tools that have blurred the lines between truth and fabrication. The surge in wartime misinformation has exposed an urgent need for stronger detection tools, experts say, as major tech platforms have largely weakened safeguards by scaling back content moderation and reducing reliance on human fact-checkers. After Iran struck Israel with barrages of missiles last week, AI-generated videos falsely claimed to show damage inflicted on Tel Aviv and Ben Gurion Airport. The videos were widely shared across Facebook, Instagram and X. Using a reverse image search, AFP's fact-checkers found that the clips were originally posted by a TikTok account that produces AI-generated content. There has been a "surge in generative AI misinformation, specifically related to the Iran-Israel conflict," Ken Jon Miyachi, founder of the Austin-based firm BitMindAI, told AFP. "These tools are being leveraged to manipulate public perception, often amplifying divisive or misleading narratives with unprecedented scale and sophistication." - 'Photo-realism' - GetReal Security, a US company focused on detecting manipulated media including AI deepfakes, also identified a wave of fabricated videos related to the Israel-Iran conflict. The company linked the visually compelling videos -- depicting apocalyptic scenes of war-damaged Israeli aircraft and buildings as well as Iranian missiles mounted on a trailer -- to Google's Veo 3 AI generator, known for hyper-realistic visuals. The Veo watermark is visible at the bottom of an online video posted by the news outlet Tehran Times, which claims to show "the moment an Iranian missile" struck Tel Aviv. "It is no surprise that as generative-AI tools continue to improve in photo-realism, they are being misused to spread misinformation and sow confusion," said Hany Farid, the co-founder of GetReal Security and a professor at the University of California, Berkeley. Farid offered one tip to spot such deepfakes: the Veo 3 videos were normally eight seconds in length or a combination of clips of a similar duration. "This eight-second limit obviously doesn't prove a video is fake, but should be a good reason to give you pause and fact-check before you re-share," he said. The falsehoods are not confined to social media. Disinformation watchdog NewsGuard has identified 51 websites that have advanced more than a dozen false claims -- ranging from AI-generated photos purporting to show mass destruction in Tel Aviv to fabricated reports of Iran capturing Israeli pilots. Sources spreading these false narratives include Iranian military-linked Telegram channels and state media sources affiliated with the Islamic Republic of Iran Broadcasting (IRIB), sanctioned by the US Treasury Department, NewsGuard said. - 'Control the narrative' - "We're seeing a flood of false claims and ordinary Iranians appear to be the core targeted audience," McKenzie Sadeghi, a researcher with NewsGuard, told AFP. Sadeghi described Iranian citizens as "trapped in a sealed information environment," where state media outlets dominate in a chaotic attempt to "control the narrative." Iran itself claimed to be a victim of tech manipulation, with local media reporting that Israel briefly hacked a state television broadcast, airing footage of women's protests and urging people to take to the streets. Adding to the information chaos were online clips lifted from war-themed video games. AFP's fact-checkers identified one such clip posted on X, which falsely claimed to show an Israeli jet being shot down by Iran. The footage bore striking similarities to the military simulation game Arma 3. Israel's military has rejected Iranian media reports claiming its fighter jets were downed over Iran as "fake news." Chatbots such as xAI's Grok, which online users are increasingly turning to for instant fact-checking, falsely identified some of the manipulated visuals as real, researchers said. "This highlights a broader crisis in today's online information landscape: the erosion of trust in digital content," BitMindAI's Miyachi said. "There is an urgent need for better detection tools, media literacy, and platform accountability to safeguard the integrity of public discourse." burs-ac/jgc
