Latest from AFP
AFP
7 hours ago
- Politics
- AFP
Video purported to show strike at Iran's Evin prison: what we know
lities escalated on June 13, when Israel launched an aerial attack on sites in Iran after again accusing Tehran of pursuing a military nuclear program. Iran responded with a series of missile and on Israel. nearly two weeks of deadly attacks, including US strikes on Iran's nuclear facilities, the countries accepted a ceasefire agreement on June 24. Iranian authorities said certain parts of Evin prison, which rights groups say holds Western detainees, political prisoners and opposition members, were A post on X claims to show a six second, black and white video appearing to be CCTV footage of an explosion at the entrance of the The video is filmed at a fixed angle, with text reading, "CAMERA 07" in the top left AFP also found the video circulating on YouTube, Facebook, Instagram and Threads with additional posts in French, Spanish and Arabic garnering thousands of views and hundreds of interactions. The video was used by multiple media outlets worldwide, including the BBC, Le Monde, NBC News, The Guardian, France24, The Times and the New York Post. But as doubts emerged around the footage it was removed from several media websites, accompanied by messages explaining uncertainties about its authenticity. "AFP's Middle East video editors saw the images on Monday but given their dubious nature decided against using them," said Louis Massie, central editor-in-chief for video. Tal Hagin, information warfare researcher for FakeReporter, an Israeli watchdog group, published a series of posts on X explaining his suspicions on June 23 (archived here). Image A screenshot of one of Tal Hagin's posts on X arguing the image was AI-generated taken June 27, 2025 "At first, I believed it to be real," he told AFP June 25, but "once I began to receive several inquiries about its validity, I looked closer." He determined the video was likely AI-generated, using a real image of the prison as a visual "That, combined with several minute errors in the footage -- door was blown off while simultaneously not blown off -- and comparing it to real aftermath footage of the strike on the prison, I concluded with a high degree it was fake." Multiple inconsistencies An AFP analysis of the video also suggested the content was AI-generated. First, most professional video surveillance cameras automatically display the date and time on footage. This is not the case in the widely shared video. Even though an incorrect or different configuration of the date and time is theoretically possible, the absence of this information is dubious at an institution as secure as Evin prison. Image Screenshots taken June 25, 2025 of two videos circulating on social media which originate from a video surveillance camera In place of the date and time, the circulating video displays in the top left corner text reading "CAMERA 07," but the country's dominant language is Farsi, not English. The text is also suspicious as it lacks a relief or any shadow. The text briefly disappears at the moment of the explosion, as if it were added during an editing process. Image Screenshots of the viral video taken June 25, 2025 The image is in black and white, while the explosion allegedly occurred during the day. Video surveillance cameras only switch to monochrome mode in low or infrared light. This monochrome processing could also be used to mask certain graphical inconsistencies. Two screenshots, taken one second apart, identify the facade of the building as the Evin prison. In the first image, we observe: A closed metal door A sign in Farsi Trees without leaves or bushes in the foreground The text "CAMERA 07" in the top left corner In the second image, an explosion seems to happen. A cloud of smoke forms in front of the door, and a flash of light appears. But many elements are suspect: No reaction from the scenery : The trees and bushes in the foreground stay perfectly still, without the slightest tremor or perceptible movement. No visible fallout : No debris, nor shockwave, is identified in the image or on the ground Suspect light : The bright halo does not seem realistic -- it seems "glued" to the facade, without any bounce or reflection on the surrounding elements. A white film appears stealthily in the shot, without a visible link to the scene -- an artifact typical of animation or poorly finalized synthesis. Image Two screenshots of the video taken at one second intervals, captured June 25, 2025 Moreover, no visible projection of debris or shockwave is identifiable between the two screenshots. e Evin prison was touched by an explosion, as seen in a video by Islamic Republic of Iran Broadcasting (archived here). But in all other images besides the purported CCTV footage, the entire building structure surrounding the door is on the ground, not only the door as in the viral Using an archival image? Another key element reinforced suspicions about the clip -- reverse image searches reveal its resemblance to an archive photo of the prison. A comparison with a photo taken in 2023 (left photo below) shows a near-perfect match to the point of view, plants, sign, trees and enclosure wall . The strong similarity suggests this image was used as a base in the process of generation or manipulation by artificial intelligence. The fact the plant elements are identical between the old photo and the purported current video further weakens the argument that the video shows an authentic scene. Image Comparison between a photo of the Evin prison from 2023 (left) and a screenshot of the viral video taken June 25, 2025 Moreover, the trees are sparce, which suggests the scene was filmed in winter. The images circulated by Iranian media officials regarding the 2025 strike on the Evin prison show lush greenery outside the walls. Image Screenshots taken by Tal Hagin on X Architectural differences were also identified when comparing the video to recent photos. The American disinformation watchdog NewsGuard has identified more than 50 sites and Telegram channels distributing deceptive content online discussing the conflict, including some related to influence operations by Iran, Russia and China. Since the start of the Iran-Israel war, AFP has found an increase in content manipulated or generated by AI. You can find these articles and other fact checks related to the conflict here.
AFP
7 hours ago
- Politics
- AFP
Canada will not enact new nationwide driving rules in July 2025
"Starting July 5,2025, big changes are coming to Canada's roads. The federal government will introduce the New Canada Driving Law 2025," claims a June 17, 2025 Facebook post. The text purports Canada is about to enforce a country-wide 30 kilometers (18.6 miles) per hour speed limit in all school zones, coming with a violation fine of Numerous other posts spread across Facebook claiming the country would be lowering the blood alcohol limit for impaired driving violations, requiring mandatory installation of advanced driver-assistance systems and forcing seniors to submit to driving evaluations. The text in many posts appeared to be copy-and-pasted, while some publications linked to articles repeating similar claims on websites with domains indicating activity Image Screenshot of a Facebook post taken June 27, 2025 Image Screenshot of a Facebook post taken June 27, 2025 posts gained traction as the Liberal government faces mounting pressure to adjust a ban on new gas-powered vehicle sales scheduled to go into effect in 2026. Other false claims about Prime Minister Mark Carney and new Canadian auto regulations were recently fact-checked by AFP. Transport Canada, the federal agency regulating transportation, also told AFP it did not have any new motor vehicle rule changes coming into effect in July 2025. "It's important to note that provinces and territories are responsible for areas such as driver licensing, vehicle registration, vehicle maintenance, insurance and enforcement of road rules," said Hicham Ayoun, a spokesman for Transport Canada in a June 19 email. The Ontario Ministry of Transportation told AFP it was not introducing any new rules such as the ones seen in the posts. "At this time there are no changes to Ontario's distracted driving law, school zones or automated vehicle oversight," said ministry spokeswoman Tanya Blazina in a June 26 email. AFP also confirmed with the province of British Columbia that the driving regulations seen in the misleading social media posts would not be coming to its jurisdiction. Searching the websites of regulations in certain provinces does not reveal any evidence of similar rules being introduced in those regions of the country (archived here, here and here) ffic Safety Act will go into effect on July 1 in Yukon territory to increase fines for some violations including failing to stop for school busses ( Read more of AFP's reporting on misinformation in Canada here.
AFP
9 hours ago
- Health
- AFP
Experts say HHS document misrepresents studies on Covid-19 vaccine
Kennedy, who has a long history of promoting vaccine misinformation, is using his role as health secretary to shake up the country's approach to immunization. He has deflected questions from lawmakers about measles vaccination, despite an outbreak that has killed three children, and misrepresented the position of European health agencies regarding vaccines against chickenpox during Congressional testimony. In late May, he circumvented the usual channels for updating vaccine recommendations and announced that the US Centers for Disease Control and Prevention (CDC), which operates under his department, would stop recommending routine Covid-19 shots for pregnant and "healthy children." Image US Secretary of Health and Human Services Robert F. Kennedy Jr. speaks during a news conference to discuss health insurance at the Department of Health and Human Services Headquarters in Washington, DC, on June 23, 2025 (AFP / SAUL LOEB) Maria Velez, an associate professor of obstetrics and gynecology at Canada's (archived here), told AFP the findings of her paper, "Miscarriage after SARS-CoV-2 vaccination: A population-based cohort study," were misinterpreted in the text (archived here). "Our study shows that SARS-CoV-2 vaccine is not associated with an increased risk of miscarriage, either for remotely vaccinated or recently vaccinated women," she said in an email on June 23. AFP reached out to HHS for comment and did not receive a response, but the department previously told other publications it included Velez's research because it showed a higher occurrence of miscarriage among vaccinated individuals. Velez said the raw data included in her study showed a slightly higher incidence of miscarriage among pregnant people who received the shot, but pointed out when the results were adjusted for other variables which could result in loss of pregnancy, her findings did not show an increased risk associated with vaccination. Additionally, a reproductive endocrinologist and infertility specialist based in New York City (archived here), told AFP her study looking at Covid-19 vaccination in people undergoing in-vitro fertilization cited in the document did not find an association between the shots and adverse stimulation or early pregnancy outcomes (archived here). "Our study provides evidence to support safety of Covid-19 vaccination in women who are trying to conceive," she said in a June 26 email. said the document misused data and incorrectly quoted findings, sowing doubt about the safety of vaccines. "The latest correspondence from HHS regarding the decision to rescind the Covid-19 vaccine recommendation for pregnant women further confirms that the decision was not made based on any new research or latest scientific evidence, Research has demonstrated that the Covid-19 vaccine is generally safe during pregnancy and a meta-analysis of 66 studies found vaccination reduced the odds of infection and hospitalization, while the most common adverse side effect was pain at the injection site (archived here and here). Children and Covid-19 vaccines The Covid-19 vaccines are estimated to have saved millions of lives (archived here). Physicians and immunology experts have continually told AFP the risks of being infected with the virus far outweigh potential, infrequent side effects from the shots (archived here). Age raises the risk of serious illness and the World Health Organization only recommends vaccination beyond an initial series for children and adolescents with comorbidities (archived here and here). According to the CDC website, the agency still recommends boosters for children who are sed and The HHS memo sent to lawmakers put a particular emphasis on myocarditis and pericarditis, inflammations of the tissue around the heart. While these are noted as possible side effects of Covid-19 vaccination, with a slightly higher prevalence of the reaction observed among younger male recipients of mRNA shots, the papers cited in the memo included a study previously featured in misleading claims debunked by AFP (archived here). The research only found the inflammatory conditions among vaccinated youth, but one of the paper's authors noted to AFP at the time that the observational study may have missed cases in unvaccinated patients which would have been picked up in a randomized trial. The study also found the cases of myocarditis and pericarditis were mild and fast resolving, while vaccination reduced hospitalization from Covid-19. Another study looking at data from the HHS run Vaccine Adverse Events Reporting System claimed a strong association between vaccination and myocarditis and death (archived here), but one of the authors has a history of spreading false information about the shots. Additionally, the journal Therapeutic Advances in Drug Safety, where the study appeared, issued an expression of concern about potential issues with the methodology and conflicts of interest (archived here). Dubious evidence The misrepresentation of studies' findings in the document fits into a larger pattern of HHS overhauling health policy while citing dubious evidence. The highly anticipated "Make America Healthy Again" report released on May 22 investigating children's health was initially published citing several sources that did not exist. It was updated, but experts said it still contained errors, including the misrepresentation of research findings. Kennedy also dismissed all 17 members of the Advisory Committee on Immunization Practices (ACIP), of being compromised by financial ties to pharmaceutical companies. He replaced them with several individuals known to spread vaccine misinformation, including controversial researcher Robert Malone, who has promoted the antiparastic drug ivermectin to treat Covid-19. The former ACIP members published an editorial in the JAMA medical journal, saying their removal and the reduction Image Robert Malone speaks during a first meeting of the CDC's Advisory Committee On Immunization Practices on June 25, 2025 in Atlanta, Georgia (GETTY IMAGES NORTH AMERICA / Elijah Nouvelage) The committee said it plans to revisit the childhood vaccine schedule and voted to bar thimerosal, a rarely used ingredient that can prevent bacterial contamination in multidose vials of influenza vaccines. ging from the anti-vaccine movement, which regularly questions shot ingredients despite no evidence of harm. Read more of AFP's reporting on health misinformation here.
AFP
12 hours ago
- Politics
- AFP
Mass explosion video is AI-generated, not US attack on Iran
"Heavy water nuclear plant at Arak. I guess Trump wasn't bluffing after all - just part of the fireworks in Iran. Expect to see more surprises," 21, 2025 Facebook post sharing the clip of a mushroom cloud detonation over a residential area. Image A screenshot of a Facebook reel taken June 26, 2025 video, which also circulated in Spanish and in French, spread as violence escalated in the Middle East, with Israel bombarding Iran and the US military striking its nuclear installations before a ceasefire was reached. attacked three Iranian sites key to Tehran's nuclear program on the night of June 21, hitting targets in the provinces of Natanz, Isfahan and the mountain-buried Fordo. The strikes added to a 12-day Israeli campaign that also targeted the country's top military brass and saw Iran retaliate by firing waves of missiles at Israel. US President Donald Trump has insisted the operation was a "spectacular military success" that "obliterated" Iran's nuclear sites, despite an intelligence assessment that raised doubts and claims from the Iranian government that it had "taken the necessary measures" to ensure the continuation of its program. Arak's heavy water reactor was attacked June 19 by Israel, not the US military, according to the Israel Defense Forces (archived here). But the video of the massive blast circulating on social media is AI-generated. A Google reverse image search uncovered an identical video posted June 18 to YouTube by the Turkey-based account "@cmlacyn" (archived here). The video's title -- as well as comments from the author -- reference AI usage. The owner of the account, Cemil Aciyan, states in his bio that "all videos on this channel are produced with artificial intelligence" (archived here). a June 20 direct message on Instagram, Aciyan confirmed to AFP: "I created all the videos on my channel with artificial intelligence." A search on Aciyan's other social media platforms yielded results for the same video on Instagram with the caption: "It's not real, I produced it with artificial intelligence" (archived here). AFP has debunked a slew of online misinformation about Iran here.
AFP
17 hours ago
- Politics
- AFP
AI fakes duel over Sara Duterte impeachment in Philippines
Neither were real. The schoolboys and elderly woman making their cases were AI creations, examples of increasingly sophisticated fakes possible with even basic online tools. "Why single out the VP?", a digitally created boy in a white school uniform asks, arguing that the case was politically motivated. The House of Representatives impeached Sara Duterte in early February on charges of graft, corruption and an alleged assassination plot against former ally and running mate President Ferdinand Marcos (archived link). But after convening as an impeachment court on June 10, the senior body immediately sent the case back to the House, questioning its constitutionality (archived link). Duterte ally Senator Ronald dela Rosa shared the video of the schoolboys -- since viewed millions of times -- praising the youths for having a "better understanding of what's happening" than their adult counterparts. The vice president's younger brother Sebastian, mayor of family stronghold Davao, said the clip proved "liberals" did not have the support of the younger generation. When the schoolboys were exposed as digital creations, the vice president and her supporters were unfazed. "There's no problem with sharing an AI video in support of me. As long as it's not being turned into a business," Duterte told reporters (archived link). "Even if it's AI... I agree with the point," said Dela Rosa, the one-time enforcer of ex-president Rodrigo Duterte's drug war. Five minutes' work The video making the case for impeachment -- also with millions of views -- depicts an elderly woman peddling fish and calling out the Senate for failing to hold a trial. "You 18 senators, when it's the poor who steal, you want them locked up immediately, no questions asked. But if it's the vice president who stole millions, you protect her fiercely," she says in Tagalog. Both clips bore a barely discernible watermark for the Google video-generation platform Veo. AFP fact-checkers also identified visual inconsistencies, such as overly smooth hair and teeth and storefronts with garbled signage. Image Screenshot of the AI-generated video with the "Veo" watermark highlighted by AFP Image Screenshot of visual details in the AI-generated image highlighted in red by AFP The man who created the fish peddler video, Bernard Senocip, 34, told AFP it took about five minutes to produce the eight-second clip. Reached via his Facebook page, Senocip defended his work in a video call, saying AI characters allowed people to express their opinions while avoiding the "harsh criticism" frequent on social media. "As long as you know your limitations and you're not misleading your viewers, I think it's fine," he said, noting that -- unlike the Facebook version -- he had placed a "created by AI" tag on the video's TikTok upload. While AFP has previously reported on websites using hot-button Philippine issues to generate cash, Senocip said his work was simply a way of expressing his political opinions. The schoolboy video's creator, the anonymous administrator of popular Facebook page Ay Grabe, declined to be interviewed but said his AI creations' opinions had been taken from real-life students (archived link). AFP, along with other media outlets, is paid by some platforms including Meta, Google and TikTok for work tackling disinformation. 'Grey area' Using AI to push viewpoints via seemingly ordinary people can make beliefs seem "more popular than they actually are", said Jose Mari Lanuza of Sigla Research Center, a non-profit organisation that studies disinformation (archived link). "In the case of the impeachment, this content fosters distrust not only towards particular lawmakers but towards the impeachment process." While some AI firms have developed measures to protect public figures, Jose Miguelito Enriquez, an associate research fellow at Nanyang Technological University, said the recent Philippine videos were a different animal (archived link). "Some AI companies like OpenAI previously committed to prevent users from generating deepfakes of 'real people', including political candidates," he said. "But... these man-on-the-street interviews represent a grey area because technically they are not using the likeness of an actual living person." Crafting realistic "humans" was also getting easier, said Dominic Ligot, founder of Data and AI Ethics PH (archived link). "Veo is only the latest in a string of rapidly evolving tools for AI media generation," he said, adding the newest version produced "smoother, more realistic motion and depth compared to earlier AI video models". Google did not reply when AFP asked if they had developed safeguards to prevent Veo from being used to push misinformation. For Ligot, guardrails around the swiftly evolving technology are a must, warning AI was increasingly being used to "influence how real people feel, pressure decision-makers and distort democratic discourse".
