Latest news with #DeepFake-O-Meter
Yahoo
20-05-2025
- Entertainment
- Yahoo
Fact Check: Beware AI-generated audio claiming Qatari leader said he regrets inviting Trump to Middle East
Claim: A video authentically shows the emir of Qatar, Sheikh Tamim bin Hamad Al Thani, saying he regrets inviting U.S. President Donald Trump to the Middle East. Rating: Shortly after U.S. President Donald Trump's visit to Qatar in May 2025, a rumor circulated online that the country's ruling emir said he regretted inviting the American commander-in-chief "to the Middle East." "He didn't come to discuss cooperation, he came to collect tributes," Qatar's leader, Sheikh Tamim bin Hamad Al Thani, allegedly said, per the clip. "Fountains, palaces, golden welcoming ceremonies, light shows, private flights. In his eyes, these grand gestures are not courtesies, but spoils of war." The video spread across social media platforms, including X, Facebook, Threads, Substack and TikTok. Snopes readers also wrote in to ask whether a Qatari or Saudi official said he regretted inviting Trump to visit; the president went to Qatar, Saudi Arabia and the United Arab Emirates during his trip across the Middle East. This clip's audio appears to be generated by artificial intelligence, according to several tools that detect AI usage. Furthermore, the video of the sheikh is either AI-generated or an edited version of a 2017 interview with CBS' "60 Minutes." Thus, we rate this claim fake. An AI detector available through "DeepFake-O-Meter," a project developed by the University of Buffalo Forensic Laboratory, rated the likelihood the audio was AI-generated as 100%. Another tool — "Deepfake Total," developed by the Fraunhofer Institute for Applied and Integrated Security — gave the audio clip a 75.6% chance of being AI-generated. Several eagle-eyed internet users noted that the clip appeared to match a 2017 interview with CBS' "60 Minutes," as seen in this YouTube video of the exchange from news outlet Al Jazeera, suggesting the clip is not AI-generated but outdated. However, when shown side-by-side, it's clear whoever made the clip either edited the CBS interview using video editing software or generated a clip through AI using the "60 Minutes" video as source material. See, for example, the smoothing of Al Thani's face in the faked clip (left) and the differing structure of his face in the legitimate interview (right) for comparison: (Substack user NEW YORK MINUTE and Al Jazeera English/CBS 60 Minutes) In the "60 Minutes" interview, the emir discusses a 2017 trip during Trump's first term, but nothing he says in the clip posted by Al Jazeera or the transcript published by CBS aligns with the AI-generated audio from May 2025. Thus, the video accompanying the AI-generated voice purporting to be the Qatari leader is either AI-generated or unrelated to Trump's 2025 visit. The clip and its claims seemed to originate with a TikTok video from user @tian128560m, per the username included in the videos circulating. That user no longer has the clip in question available on their profile, but several of their videos appear to also be misinformation, including one which falsely claims Barron Trump, son of U.S. President Donald Trump, dropped out of New York University, despite numerous reports that he finished his first year at NYU in May. We have reached out to the TikTok user and will update this story if we receive more information. Snopes has previously fact-checked many claims about Qatar and its leaders; see our collection of fact-checks on the country here. Al Jazeera English. "🇶🇦 Qatar Emir: Our Sovereignty Is a Red Line." YouTube, 29 Oct. 2017, Accessed 19 May 2025. "DeepFake-O-Meter." University of Buffalo , 2025, Accessed 19 May 2025. "DEEPFAKETOTAL." Cognitive Security Technologies (CST) department of Fraunhofer AISEC, Accessed 19 May 2025. Rose, Charlie. "Qatar's Emir Stands Defiant in Face of Blockade." CBS News, 29 Oct. 2017, Accessed 19 May 2025.
Yahoo
24-03-2025
- Politics
- Yahoo
Fact Check: No evidence clip of Vance criticizing Musk is authentic
Claim: An audio clip authentically features U.S. Vice President JD Vance criticizing tech billionaire Elon Musk. Rating: On March 23, 2025, an audio clip circulated online in which U.S. Vice President JD Vance appeared to criticize tech billionaire Elon Musk, an adviser to President Donald Trump. Social media users posted the clip across YouTube (archived), Facebook (archived), X (archived), Threads (archived), Instagram (archived), Reddit (archived) and Bluesky (archived). According to the posts, Vance said: Everything that he's doing is getting criticized in the media. And he says that he's helping, and he's not. He's making us look bad. He's making me look bad. And I'll tell you this, and he wouldn't like it if I said it, but he's not even an American. He's from South Africa and he's cosplaying as this great American leader in a room that has the portraits of some of the greatest men that ever led this country, and he has the audacity to act like he is an elected official. I am an elected official, I am the important one in this situation, not him. So if he wants to tank the economy and his cars maybe that's what he deserves. However, on March 24, Vance's communications director William Martin said (archived) on X that the audio was "100% fake and most certainly not the Vice President." The origins of the audio were unclear, but several artificial intelligence detectors said the audio was likely not authentic. There were no reports from reputable sources about the audio or who might have recorded it. As such, we have rated this video fake. We reached out to the White House for clarification on Martin's statement. We also reached out to two early TikTok posters, whose videos were widely shared by other social media users, about where they got the audio and await their replies. One of the TikTok profiles that posted an early version of the audio told a commenter who asked how they knew the audio was authentic: " it was AI..I would label as such...I don't Both profiles that posted early versions of the audio have posted other political content with an anti-Trump lean. We passed the audio through Hiya's Deepfake Voice Detector and the University at Buffalo Media Lab's DeepFake-O-Meter. (Deepfake audio is audio generated by AI to sound like a real person, in this case Vance.) According to Hiya, audio samples from the beginning, middle and end of the 55-second clip scored from 32 to 52 of 100 points, with 100 being a 100% authentic voice. The beginning of the clip was "likely" to be fake, according to the the detector scoring 32/100. The detector was uncertain about the other samples. (Screenshots from Hiya Deepfake Voice Detector) DeepFake-O-Meter's seven detectors said the audio was between 70.2% and 100% likely to be AI-generated. Whereas Hiya took a short sample of several seconds, DeepFake-O-Meter analyzed the entire audio clip. (Screenshot from University at Buffalo Media Lab's DeepFake-o-meter) Though all the available evidence points to the recording not being real, it does play on swirling rumors that Vance felt pushed aside by Musk's prominence in the Trump administration. Since Trump's inauguration, Musk has appeared alongside Trump at least once for a news conference. Vance has not. Musk also has met with several foreign leaders alongside the president, a role usually filled by the VP. In a March 14 interview with NBC News, Vance conditionally supported Musk's work with the Department of Government Efficiency. Vance said he was "accepting of mistakes" but "I also think you have to quickly correct those mistakes." Allison, Natalie, and Jacqueline Alemany. "In Trump's Crowded White House, JD Vance Finds a Role behind the Scenes." Washington Post, 10 Mar. 2025, @amouthful_of_cherries. "Whoever's Leaking All These Recordings Is Doing the Lord's Work 🙏🙌✨." Instagram, 24 Mar. 2025, @christinatalkstea. "#teawithchristina." TikTok, 23 Mar. 2025, "Leaked Audio of JD Vance in a Jealous Rant…." Bluesky, 24 Mar. 2025, Deepfake Voice Detector. Accessed 24 Mar. 2025. @ejacobson. "This Leaked Audio of JD Vance Shading Elon Musk Is Funny for Two Reasons..." X, 24 Mar. 2025, "Elon Musk's Meetings with Foreign Leaders alongside Trump Alarm Washington Insiders." NBC News, 20 Dec. 2024, Fox 7 Austin. "Elon Musk at the White House: FULL NEWS CONFERENCE." YouTube, 12 Feb. 2025, @josey6529. "#italy #france." TikTok, 23 Mar. 2025, @occupydemocrats. "THIS IS HUGE! Leaked Audio by JD Vance Criticizing #ElonMusk." Threads, 24 Mar. 2025, The Political Insider. "'He Is Making Us Look Bad'." Facebook, 24 Mar. 2025, @TonyMichaelsPodcast. "Leaked Audio of JD Vance on Elon Musk 😳!!!" YouTube, 23 Mar. 2025, UB Media Forensics Lab. "DEEPFAKE-O-METER." DEEPFAKE-O-METER, u/Shenanie-Probs. "Leaked JD Vance Audio on Elon Musk." Reddit, 23 Mar. 2025, "Vance Discusses Elon Musk's 'mistakes' and 'Incremental Progress' on the Economy in NBC News Interview." NBC News, 14 Mar. 2025, @wsmartin218. "This Audio Is 100% Fake and Most Certainly Not the Vice President. ." X, 24 Mar. 2025,
Yahoo
17-02-2025
- Yahoo
AI image passed off as real photo of firemen kneeling in prayer during LA wildfires
The image, which appears to show a circle of firemen kneeling in prayer while a large fire blazes behind them, was shared on Facebook on February 10. "Recent wildfire in Los Angeles in the United States. It shows firemen kneeling in prayer after they did their best to put out the fire but failed," reads the Korean-language caption. The blazes that erupted around Los Angeles on January 7 during hurricane-strength wind storms devoured 40,000 acres (16,000 hectares) and killed about 30 people, before finally being fully contained after three weeks (archived link). The Palisades and Eaton fires in Southern California's Los Angeles County were the most destructive in the history of the second-largest US city, burning more than 37,000 acres (150 square kilometres) and over 10,000 homes, causing damage estimated to cost hundreds of billions of dollars. Facebook users here, here and here shared the image as a genuine one from the fires, but it was in fact generated with AI. Google reverse image searches led to several older instances of the image shared online that featured a watermark on the bottom-right corner that read "picture from AI" (archived links here and here). This disclaimer was cropped out of the false posts. The "About this image" feature on Google's reverse image search also indicated that the picture was "Made with Google AI". Google's DeepMind AI lab launched the SynthID feature in 2023, a tool for digitally watermarking and identifying images made with Google AI (archived link). "Users can identify if an image, or part of an image, was generated by Google's AI tools through the About this image feature in Search or Chrome," its website says (archived link). AI experts also pointed out inconsistencies in the image that indicated it was generated with AI. Siwei Lyu, director of the University at Buffalo's Media Forensics Lab in the United States, told AFP that his team's DeepFake-O-Meter tool determined the image was "very likely to be generated by AI" (archived links here and here). The image contained visual inconsistencies -- such as irregular text on the firefighters' uniforms, inconsistent details on the fire trucks and road markings that changed width and colour -- indicating it was not a genuine photo. Shu Hu, head of Purdue University's Purdue Machine Learning and Media Forensics Lab in the United States, also said the image was likely AI-generated, with the most apparent artefact being that the "human shadows are indistinguishable, even under strong backlighting" (archived link). AFP has debunked other AI videos and images of the Los Angeles fires and another of a house that purportedly survived the inferno.
