{"id":1834208,"date":"2026-03-19T08:23:56","date_gmt":"2026-03-19T05:23:56","guid":{"rendered":"https:\/\/analyse.optim.biz\/?p=1834208"},"modified":"2026-03-19T08:23:56","modified_gmt":"2026-03-19T05:23:56","slug":"fact-check-how-fake-images-from-iran-misled-media-outlets","status":"publish","type":"post","link":"https:\/\/analyse.optim.biz\/?p=1834208","title":{"rendered":"Fact check: How fake images from Iran misled media outlets"},"content":{"rendered":"

[analyse_image type=”featured” src=”https:\/\/static.dw.com\/image\/76387461_6.jpg”]<\/p>\n

\n
\n
\n
https:\/\/p.dw.com\/p\/5AY7S<\/span><\/div>\n<\/section>\n
After journalists suspected that several images provided by the news agency SalamPix were AI-generated, media houses looked through their content and found many examplesImage: Salampix<\/small><\/figcaption><\/figure>\n
\n

Manipulated or recycled photos and videos\u00a0<\/span>frequently\u00a0circulate during wars and crises \u2014 sometimes to deceive, sometimes simply to generate clicks.<\/p>\n

During the ongoing\u00a0US\u2011Israel war with Iran<\/a>, the problem has reached a new level.\u00a0Photo agencies themselves\u00a0were supplied with manipulated\u00a0or\u00a0fake\u00a0images, which then ended up in newsrooms across Europe.\u00a0<\/p>\n

Some of the images appear to have been generated with artificial intelligence, while others were digitally altered by humans.<\/span>\u00a0Here is the story, what we learned and\u00a0how you\u00a0can\u00a0identify\u00a0AI<\/a> fakes.\u00a0<\/p>\n

The\u00a0<\/span>SalamPix\u00a0saga\u00a0<\/h2>\n

In early March,\u00a0<\/span>Dutch media reported<\/a><\/span><\/span>\u00a0that ANP (<\/span>Algemeen\u00a0Nederlands\u00a0Persbureau), the country’s largest news agency, removed\u00a0roughly 1,000\u00a0Iran\u2011related photos from its database after suspecting that some had been manipulated with AI.\u00a0<\/p>\n

Two days later, the Dutch branch of the media network\u00a0<\/span>RTL reported<\/a><\/span><\/span>\u00a0that its news service, RTL\u00a0<\/span>Nieuws, had unknowingly used three of these images on its website and in its app. After ANP alerted RTL that the photos were AI\u2011generated, RTL removed them and published a detailed explanation\u00a0identifying\u00a0which images had been taken down and why.\u00a0<\/p>\n

\"UGC
When RTL was alerted that photos it had used were AI\u2011generated, it removed them and published a explanation identifying which images had been taken down and why Image: rtl<\/small><\/figcaption><\/figure>\n

Shortly afterward, Germ<\/span>an\u00a0weekly news magazine Der Spiegel<\/em> acknowledged that it had also\u00a0used an AI\u2011manipulated image<\/a><\/span><\/span>\u00a0in its coverage before realizing it was fake.<\/span>\u00a0<\/p>\n

In both cases, reputable news agencies had supplied the images. RTL received them through ANP, while Der Spiegel<\/em> obtained them via dpa Picture Alliance,\u00a0<\/span>ddp\u00a0and\u00a0Imago Images,\u00a0all of which had sourced the material from the French agency Abaca Press.<\/p>\n

The images were\u00a0<\/span>ultimately traced\u00a0back to an Iranian agency,\u00a0SalamPix. According to Der Spiegel<\/em>,\u00a0SalamPix\u00a0provided the photos to Abaca Press, which then distributed them to the databases of multiple international agencies and, ultimately, to newsrooms.\u00a0<\/p>\n

News agencies<\/span>\u00a0normally collect\u00a0media\u00a0content, including pictures,\u00a0from around the world and then\u00a0sell and\u00a0distribute\u00a0the information\u00a0to\u00a0newspapers, TV\u00a0and radio stations,\u00a0which in turn use it\u00a0as part of their own coverage.<\/p>\n

In response to the revelations of AI-manipulated images, many photo agencies have blocked\u00a0<\/span>SalamPix\u00a0or issued\u00a0so-called\u00a0“kill notices,”\u00a0instructing media\u00a0clients\u00a0they\u00a0work with\u00a0to remove\u00a0SalamPix\u00a0images from publication.<\/p>\n

How the photo agencies were fooled<\/span>\u00a0<\/h2>\n

In Germany, the “agency privilege”<\/span>\u00a0is\u00a0<\/span>a legal concept that\u00a0generally\u00a0allows\u00a0media outlets to rely on the authenticity of verified text, image\u00a0and video material delivered by\u00a0news agencies with whom they work.\u00a0Even a global broadcaster like Deutsche Welle (DW) routinely relies on external agencies to cover global events.\u00a0<\/p>\n

But as AI\u2011generated and AI\u2011manipulated content becomes more sophisticated and prolific, distinguishing real images from fabricated ones is increasingly difficult. The challenge is not only technological\u00a0but\u00a0also logistical. During fast-moving breaking news<\/span> situations, journalists and agencies must sift through massive volumes of visuals at high speed<\/span>.\u00a0Since the start of 2026, DW has received an average of 1<\/span>40,000 images per day from agencies.\u00a0<\/span>\u00a0<\/p>\n

“<\/span>Transparency is one of our highest priorities. Whenever we show AI\u2011generated content, it must be clearly and unmistakably identifiable as such,”\u00a0explains DW Editor\u2011in\u2011Chief\u00a0Mathias Stamm.\u00a0“And if we make a mistake \u2014 as in the case of using images from the agency\u00a0SalamPix\u00a0\u2014 we acknowledge it and remain transparent.”\u00a0<\/p>\n

Examples used and\u00a0<\/span>identified\u00a0by DW\u00a0<\/h2>\n

After the\u00a0<\/span>first<\/span>\u00a0reports\u00a0<\/span>about\u00a0<\/span>SalamPix<\/span>, DW reviewed its own coverage and found that it, too, had used\u00a0<\/span>its<\/span>\u00a0images.<\/span>\u00a0DW subsequently removed all SalamPix images from its publications and\u00a0issued correction statements<\/a><\/span><\/span>\u00a0under each changed article\u00a0<\/span>to explain the altered content.\u00a0\u00a0<\/p>\n

\"An
This image from February, allegedly showing the aftermath of a missile’s impact in Tehran, is actually AI-generatedImage: SalamPix\/ABACA<\/small><\/figcaption><\/figure>\n

One example is a\u00a0<\/span>seemingly realistic\u00a0street scene\u00a0depicting an alleged\u00a0missile strike in Tehran, showing yellow cars (possibly taxis) in the foreground, buildings in the background and smoke rising behind them.\u00a0It was shared through the agencies in\u00a02026.\u00a0<\/p>\n

On closer inspection, however, several unmistakable AI glitches<\/span><\/span>\u00a0become obvious.\u00a0<\/p>\n

\"Detail
When you zoom in, it becomes obvious that the text is gibberish \u2014 a typical AI glitchImage: SalamPix\/ABACA<\/small><\/figcaption><\/figure>\n

Most notably, the writing on a wall and on a car looks like text \u2014 until examined more closely. Upon zooming in, it clearly\u00a0does not resemble Farsi or Arabic script. In fact, it is not any language at all, but nonsensical pseudo\u2011text, a common glitch in AI\u2011generated imagery.<\/span>\u00a0<\/p>\n

\"Detail
Another typical AI glitch is that usually straight surfaces, like walls and windows, seem to bulge out like a bubble, as seen hereImage: SalamPix\/ABACA<\/small><\/figcaption><\/figure>\n

Another typical mistake in AI-generated images is the presence of oddly shaped\u00a0buildings and objects\u00a0that defy logic. In this case, we also found an oddly shap<\/span>ed car window and bulging walls and windows of the\u00a0building that\u00a0is\u00a0above the yellow car in the\u00a0<\/span>center\u00a0of the picture.\u00a0<\/p>\n

\"Detail
Zooming in on cars and buses can also be revealing, as AI systems frequently produce vehicles that don’t exist or contain unrealistic, illogical detailsImage: SalamPix\/ABACA<\/small><\/figcaption><\/figure>\n

If you zoom in on the car and bus in the lower left corner of the\u00a0<\/span>image,\u00a0<\/span>you’ll\u00a0also see that they are oddly shaped and do not correspond to any existing models.<\/span><\/p>\n

\"AI-generated
Another image circulated in January that allegedly showed security forces shooting at protesters; distributed by SalamPix, it also turned out to be AI\u2011generatedImage: SalamPix\/ABACA<\/small><\/figcaption><\/figure>\n

This is another example<\/span>\u00a0of an AI-generated image. It\u00a0<\/span>portrays a\u00a0man dressed all in black\u00a0with a weapon in his hand.\u00a0The caption provided read: “Tehran Armed Iranian security forces opening fire to suppress protesters during demonstrations on January 8, 2026.”\u00a0<\/p>\n

\"Detail
If you zoom in on the feet of the man, you can notice that both are different not only in size but also in shape.Image: SalamPix\/ABACA<\/small><\/figcaption><\/figure>\n

Again, we see <\/span>AI<\/span>\u00a0glitches<\/span>: The two shoes don<\/span>‘<\/span>t match;\u00a0the person\u00a0<\/span>seems to have\u00a0two completely different feet and shoes. The shadow of the person (especially the right hand, above the gun) does not correspond to the\u00a0actual body part.<\/p>\n

\"Detail
Look closely at the right hand: a chunk between the thumb and fingers seems to be missingImage: SalamPix\/ABACA<\/small><\/figcaption><\/figure>\n

The same hand\u00a0also appears anatomically incorrect, with\u00a0a chunk seemingly\u00a0missing between the\u00a0thumb\u00a0and fingers.\u00a0\u00a0<\/p>\n

DW Fact check also<\/span>\u00a0looked through\u00a0older images provided by\u00a0SalamPix\u00a0and,\u00a0like\u00a0the\u00a0S\u00fcddeutsche\u00a0Zeitung<\/a><\/span><\/span>,\u00a0<\/span>one of Germany’s largest\u00a0<\/span>daily newspapers,\u00a0found\u00a0irregularities in images published.\u00a0\u00a0<\/p>\n

One image DW analyzed is the one at the top of the article<\/span>, allegedly showing\u00a0Iranian protesters clashing with security forces during a\u00a0protest against\u00a0the Iranian regime in Mahabad city in November 2022.\u00a0<\/p>\n

The image shows\u00a0<\/span>typical AI errors of that time: the hands,\u00a0especially the fingers, of several people in the image are deformed or look like they are made of wood. The windows of the building on the left\u00a0don’t\u00a0align at\u00a0the same angle. And the\u00a0face of the person at the far right of the image is distorted.<\/p>\n

How to spot AI\u2011generated or manipulated images<\/span>\u00a0<\/h2>\n

As AI tools improve, fabricated visuals are becoming\u00a0harder to distinguish from real ones. This means that both everyday users and professional journalists are at risk of being misled \u2014 as this case\u00a0<\/span>demonstrates.\u00a0<\/p>\n

Media organizations, including DW, are investing heavily in training staff to detect and debunk AI manipulation.\u00a0<\/span>DW’s Fact check<\/a>\u00a0team also produces\u00a0<\/span>media literacy content<\/a><\/span><\/span>\u00a0to help audiences understand how to recognize misleading images and videos.<\/span>\u00a0<\/p>\n

Edited by: Thomas Sparrow, Joscha Weber, Cristina Burack<\/em><\/p>\n<\/div>\n<\/div>\n<\/article>\n

\n

Manipulated or recycled photos and videos\u00a0<\/span>frequently\u00a0circulate during wars and crises \u2014 sometimes to deceive, sometimes simply to generate clicks.<\/p>\n

During the ongoing\u00a0US\u2011Israel war with Iran<\/a>, the problem has reached a new level.\u00a0Photo agencies themselves\u00a0were supplied with manipulated\u00a0or\u00a0fake\u00a0images, which then ended up in newsrooms across Europe.\u00a0<\/p>\n

Some of the images appear to have been generated with artificial intelligence, while others were digitally altered by humans.<\/span>\u00a0Here is the story, what we learned and\u00a0how you\u00a0can\u00a0identify\u00a0AI<\/a> fakes.\u00a0<\/p>\n

The\u00a0<\/span>SalamPix\u00a0saga\u00a0<\/h2>\n

In early March,\u00a0<\/span>Dutch media reported<\/a><\/span><\/span>\u00a0that ANP (<\/span>Algemeen\u00a0Nederlands\u00a0Persbureau), the country’s largest news agency, removed\u00a0roughly 1,000\u00a0Iran\u2011related photos from its database after suspecting that some had been manipulated with AI.\u00a0<\/p>\n

Two days later, the Dutch branch of the media network\u00a0<\/span>RTL reported<\/a><\/span><\/span>\u00a0that its news service, RTL\u00a0<\/span>Nieuws, had unknowingly used three of these images on its website and in its app. After ANP alerted RTL that the photos were AI\u2011generated, RTL removed them and published a detailed explanation\u00a0identifying\u00a0which images had been taken down and why.\u00a0<\/p>\n

\"UGC
When RTL was alerted that photos it had used were AI\u2011generated, it removed them and published a explanation identifying which images had been taken down and why Image: rtl<\/small><\/figcaption><\/figure>\n

Shortly afterward, Germ<\/span>an\u00a0weekly news magazine Der Spiegel<\/em> acknowledged that it had also\u00a0used an AI\u2011manipulated image<\/a><\/span><\/span>\u00a0in its coverage before realizing it was fake.<\/span>\u00a0<\/p>\n

In both cases, reputable news agencies had supplied the images. RTL received them through ANP, while Der Spiegel<\/em> obtained them via dpa Picture Alliance,\u00a0<\/span>ddp\u00a0and\u00a0Imago Images,\u00a0all of which had sourced the material from the French agency Abaca Press.<\/p>\n

The images were\u00a0<\/span>ultimately traced\u00a0back to an Iranian agency,\u00a0SalamPix. According to Der Spiegel<\/em>,\u00a0SalamPix\u00a0provided the photos to Abaca Press, which then distributed them to the databases of multiple international agencies and, ultimately, to newsrooms.\u00a0<\/p>\n

News agencies<\/span>\u00a0normally collect\u00a0media\u00a0content, including pictures,\u00a0from around the world and then\u00a0sell and\u00a0distribute\u00a0the information\u00a0to\u00a0newspapers, TV\u00a0and radio stations,\u00a0which in turn use it\u00a0as part of their own coverage.<\/p>\n

In response to the revelations of AI-manipulated images, many photo agencies have blocked\u00a0<\/span>SalamPix\u00a0or issued\u00a0so-called\u00a0“kill notices,”\u00a0instructing media\u00a0clients\u00a0they\u00a0work with\u00a0to remove\u00a0SalamPix\u00a0images from publication.<\/p>\n

How the photo agencies were fooled<\/span>\u00a0<\/h2>\n

In Germany, the “agency privilege”<\/span>\u00a0is\u00a0<\/span>a legal concept that\u00a0generally\u00a0allows\u00a0media outlets to rely on the authenticity of verified text, image\u00a0and video material delivered by\u00a0news agencies with whom they work.\u00a0Even a global broadcaster like Deutsche Welle (DW) routinely relies on external agencies to cover global events.\u00a0<\/p>\n

But as AI\u2011generated and AI\u2011manipulated content becomes more sophisticated and prolific, distinguishing real images from fabricated ones is increasingly difficult. The challenge is not only technological\u00a0but\u00a0also logistical. During fast-moving breaking news<\/span> situations, journalists and agencies must sift through massive volumes of visuals at high speed<\/span>.\u00a0Since the start of 2026, DW has received an average of 1<\/span>40,000 images per day from agencies.\u00a0<\/span>\u00a0<\/p>\n

“<\/span>Transparency is one of our highest priorities. Whenever we show AI\u2011generated content, it must be clearly and unmistakably identifiable as such,”\u00a0explains DW Editor\u2011in\u2011Chief\u00a0Mathias Stamm.\u00a0“And if we make a mistake \u2014 as in the case of using images from the agency\u00a0SalamPix\u00a0\u2014 we acknowledge it and remain transparent.”\u00a0<\/p>\n

Examples used and\u00a0<\/span>identified\u00a0by DW\u00a0<\/h2>\n

After the\u00a0<\/span>first<\/span>\u00a0reports\u00a0<\/span>about\u00a0<\/span>SalamPix<\/span>, DW reviewed its own coverage and found that it, too, had used\u00a0<\/span>its<\/span>\u00a0images.<\/span>\u00a0DW subsequently removed all SalamPix images from its publications and\u00a0issued correction statements<\/a><\/span><\/span>\u00a0under each changed article\u00a0<\/span>to explain the altered content.\u00a0\u00a0<\/p>\n

\"An
This image from February, allegedly showing the aftermath of a missile’s impact in Tehran, is actually AI-generatedImage: SalamPix\/ABACA<\/small><\/figcaption><\/figure>\n

One example is a\u00a0<\/span>seemingly realistic\u00a0street scene\u00a0depicting an alleged\u00a0missile strike in Tehran, showing yellow cars (possibly taxis) in the foreground, buildings in the background and smoke rising behind them.\u00a0It was shared through the agencies in\u00a02026.\u00a0<\/p>\n

On closer inspection, however, several unmistakable AI glitches<\/span><\/span>\u00a0become obvious.\u00a0<\/p>\n

\"Detail
When you zoom in, it becomes obvious that the text is gibberish \u2014 a typical AI glitchImage: SalamPix\/ABACA<\/small><\/figcaption><\/figure>\n

Most notably, the writing on a wall and on a car looks like text \u2014 until examined more closely. Upon zooming in, it clearly\u00a0does not resemble Farsi or Arabic script. In fact, it is not any language at all, but nonsensical pseudo\u2011text, a common glitch in AI\u2011generated imagery.<\/span>\u00a0<\/p>\n

\"Detail
Another typical AI glitch is that usually straight surfaces, like walls and windows, seem to bulge out like a bubble, as seen hereImage: SalamPix\/ABACA<\/small><\/figcaption><\/figure>\n

Another typical mistake in AI-generated images is the presence of oddly shaped\u00a0buildings and objects\u00a0that defy logic. In this case, we also found an oddly shap<\/span>ed car window and bulging walls and windows of the\u00a0building that\u00a0is\u00a0above the yellow car in the\u00a0<\/span>center\u00a0of the picture.\u00a0<\/p>\n

\"Detail
Zooming in on cars and buses can also be revealing, as AI systems frequently produce vehicles that don’t exist or contain unrealistic, illogical detailsImage: SalamPix\/ABACA<\/small><\/figcaption><\/figure>\n

If you zoom in on the car and bus in the lower left corner of the\u00a0<\/span>image,\u00a0<\/span>you’ll\u00a0also see that they are oddly shaped and do not correspond to any existing models.<\/span><\/p>\n

\"AI-generated
Another image circulated in January that allegedly showed security forces shooting at protesters; distributed by SalamPix, it also turned out to be AI\u2011generatedImage: SalamPix\/ABACA<\/small><\/figcaption><\/figure>\n

This is another example<\/span>\u00a0of an AI-generated image. It\u00a0<\/span>portrays a\u00a0man dressed all in black\u00a0with a weapon in his hand.\u00a0The caption provided read: “Tehran Armed Iranian security forces opening fire to suppress protesters during demonstrations on January 8, 2026.”\u00a0<\/p>\n

\"Detail
If you zoom in on the feet of the man, you can notice that both are different not only in size but also in shape.Image: SalamPix\/ABACA<\/small><\/figcaption><\/figure>\n

Again, we see <\/span>AI<\/span>\u00a0glitches<\/span>: The two shoes don<\/span>‘<\/span>t match;\u00a0the person\u00a0<\/span>seems to have\u00a0two completely different feet and shoes. The shadow of the person (especially the right hand, above the gun) does not correspond to the\u00a0actual body part.<\/p>\n

\"Detail
Look closely at the right hand: a chunk between the thumb and fingers seems to be missingImage: SalamPix\/ABACA<\/small><\/figcaption><\/figure>\n

The same hand\u00a0also appears anatomically incorrect, with\u00a0a chunk seemingly\u00a0missing between the\u00a0thumb\u00a0and fingers.\u00a0\u00a0<\/p>\n

DW Fact check also<\/span>\u00a0looked through\u00a0older images provided by\u00a0SalamPix\u00a0and,\u00a0like\u00a0the\u00a0S\u00fcddeutsche\u00a0Zeitung<\/a><\/span><\/span>,\u00a0<\/span>one of Germany’s largest\u00a0<\/span>daily newspapers,\u00a0found\u00a0irregularities in images published.\u00a0\u00a0<\/p>\n

One image DW analyzed is the one at the top of the article<\/span>, allegedly showing\u00a0Iranian protesters clashing with security forces during a\u00a0protest against\u00a0the Iranian regime in Mahabad city in November 2022.\u00a0<\/p>\n

The image shows\u00a0<\/span>typical AI errors of that time: the hands,\u00a0especially the fingers, of several people in the image are deformed or look like they are made of wood. The windows of the building on the left\u00a0don’t\u00a0align at\u00a0the same angle. And the\u00a0face of the person at the far right of the image is distorted.<\/p>\n

How to spot AI\u2011generated or manipulated images<\/span>\u00a0<\/h2>\n

As AI tools improve, fabricated visuals are becoming\u00a0harder to distinguish from real ones. This means that both everyday users and professional journalists are at risk of being misled \u2014 as this case\u00a0<\/span>demonstrates.\u00a0<\/p>\n

Media organizations, including DW, are investing heavily in training staff to detect and debunk AI manipulation.\u00a0<\/span>DW’s Fact check<\/a>\u00a0team also produces\u00a0<\/span>media literacy content<\/a><\/span><\/span>\u00a0to help audiences understand how to recognize misleading images and videos.<\/span>\u00a0<\/p>\n

Edited by: Thomas Sparrow, Joscha Weber, Cristina Burack<\/em><\/p>\n<\/div>\n

Manipulated or recycled photos and videos\u00a0<\/span>frequently\u00a0circulate during wars and crises \u2014 sometimes to deceive, sometimes simply to generate clicks.<\/p>\n

During the ongoing\u00a0US\u2011Israel war with Iran<\/a>, the problem has reached a new level.\u00a0Photo agencies themselves\u00a0were supplied with manipulated\u00a0or\u00a0fake\u00a0images, which then ended up in newsrooms across Europe.\u00a0<\/p>\n

Some of the images appear to have been generated with artificial intelligence, while others were digitally altered by humans.<\/span>\u00a0Here is the story, what we learned and\u00a0how you\u00a0can\u00a0identify\u00a0AI<\/a> fakes.\u00a0<\/p>\n

In early March,\u00a0<\/span>Dutch media reported<\/a><\/span><\/span>\u00a0that ANP (<\/span>Algemeen\u00a0Nederlands\u00a0Persbureau), the country’s largest news agency, removed\u00a0roughly 1,000\u00a0Iran\u2011related photos from its database after suspecting that some had been manipulated with AI.\u00a0<\/p>\n

Two days later, the Dutch branch of the media network\u00a0<\/span>RTL reported<\/a><\/span><\/span>\u00a0that its news service, RTL\u00a0<\/span>Nieuws, had unknowingly used three of these images on its website and in its app. After ANP alerted RTL that the photos were AI\u2011generated, RTL removed them and published a detailed explanation\u00a0identifying\u00a0which images had been taken down and why.\u00a0<\/p>\n

Shortly afterward, Germ<\/span>an\u00a0weekly news magazine Der Spiegel<\/em> acknowledged that it had also\u00a0used an AI\u2011manipulated image<\/a><\/span><\/span>\u00a0in its coverage before realizing it was fake.<\/span>\u00a0<\/p>\n

In both cases, reputable news agencies had supplied the images. RTL received them through ANP, while Der Spiegel<\/em> obtained them via dpa Picture Alliance,\u00a0<\/span>ddp\u00a0and\u00a0Imago Images,\u00a0all of which had sourced the material from the French agency Abaca Press.<\/p>\n

The images were\u00a0<\/span>ultimately traced\u00a0back to an Iranian agency,\u00a0SalamPix. According to Der Spiegel<\/em>,\u00a0SalamPix\u00a0provided the photos to Abaca Press, which then distributed them to the databases of multiple international agencies and, ultimately, to newsrooms.\u00a0<\/p>\n

News agencies<\/span>\u00a0normally collect\u00a0media\u00a0content, including pictures,\u00a0from around the world and then\u00a0sell and\u00a0distribute\u00a0the information\u00a0to\u00a0newspapers, TV\u00a0and radio stations,\u00a0which in turn use it\u00a0as part of their own coverage.<\/p>\n

In response to the revelations of AI-manipulated images, many photo agencies have blocked\u00a0<\/span>SalamPix\u00a0or issued\u00a0so-called\u00a0“kill notices,”\u00a0instructing media\u00a0clients\u00a0they\u00a0work with\u00a0to remove\u00a0SalamPix\u00a0images from publication.<\/p>\n

In Germany, the “agency privilege”<\/span>\u00a0is\u00a0<\/span>a legal concept that\u00a0generally\u00a0allows\u00a0media outlets to rely on the authenticity of verified text, image\u00a0and video material delivered by\u00a0news agencies with whom they work.\u00a0Even a global broadcaster like Deutsche Welle (DW) routinely relies on external agencies to cover global events.\u00a0<\/p>\n

But as AI\u2011generated and AI\u2011manipulated content becomes more sophisticated and prolific, distinguishing real images from fabricated ones is increasingly difficult. The challenge is not only technological\u00a0but\u00a0also logistical. During fast-moving breaking news<\/span> situations, journalists and agencies must sift through massive volumes of visuals at high speed<\/span>.\u00a0Since the start of 2026, DW has received an average of 1<\/span>40,000 images per day from agencies.\u00a0<\/span>\u00a0<\/p>\n

“<\/span>Transparency is one of our highest priorities. Whenever we show AI\u2011generated content, it must be clearly and unmistakably identifiable as such,”\u00a0explains DW Editor\u2011in\u2011Chief\u00a0Mathias Stamm.\u00a0“And if we make a mistake \u2014 as in the case of using images from the agency\u00a0SalamPix\u00a0\u2014 we acknowledge it and remain transparent.”\u00a0<\/p>\n

After the\u00a0<\/span>first<\/span>\u00a0reports\u00a0<\/span>about\u00a0<\/span>SalamPix<\/span>, DW reviewed its own coverage and found that it, too, had used\u00a0<\/span>its<\/span>\u00a0images.<\/span>\u00a0DW subsequently removed all SalamPix images from its publications and\u00a0issued correction statements<\/a><\/span><\/span>\u00a0under each changed article\u00a0<\/span>to explain the altered content.\u00a0\u00a0<\/p>\n

One example is a\u00a0<\/span>seemingly realistic\u00a0street scene\u00a0depicting an alleged\u00a0missile strike in Tehran, showing yellow cars (possibly taxis) in the foreground, buildings in the background and smoke rising behind them.\u00a0It was shared through the agencies in\u00a02026.\u00a0<\/p>\n

On closer inspection, however, several unmistakable AI glitches<\/span><\/span>\u00a0become obvious.\u00a0<\/p>\n

Most notably, the writing on a wall and on a car looks like text \u2014 until examined more closely. Upon zooming in, it clearly\u00a0does not resemble Farsi or Arabic script. In fact, it is not any language at all, but nonsensical pseudo\u2011text, a common glitch in AI\u2011generated imagery.<\/span>\u00a0<\/p>\n

Another typical mistake in AI-generated images is the presence of oddly shaped\u00a0buildings and objects\u00a0that defy logic. In this case, we also found an oddly shap<\/span>ed car window and bulging walls and windows of the\u00a0building that\u00a0is\u00a0above the yellow car in the\u00a0<\/span>center\u00a0of the picture.\u00a0<\/p>\n

If you zoom in on the car and bus in the lower left corner of the\u00a0<\/span>image,\u00a0<\/span>you’ll\u00a0also see that they are oddly shaped and do not correspond to any existing models.<\/span><\/p>\n

This is another example<\/span>\u00a0of an AI-generated image. It\u00a0<\/span>portrays a\u00a0man dressed all in black\u00a0with a weapon in his hand.\u00a0The caption provided read: “Tehran Armed Iranian security forces opening fire to suppress protesters during demonstrations on January 8, 2026.”\u00a0<\/p>\n

Again, we see <\/span>AI<\/span>\u00a0glitches<\/span>: The two shoes don<\/span>‘<\/span>t match;\u00a0the person\u00a0<\/span>seems to have\u00a0two completely different feet and shoes. The shadow of the person (especially the right hand, above the gun) does not correspond to the\u00a0actual body part.<\/p>\n

The same hand\u00a0also appears anatomically incorrect, with\u00a0a chunk seemingly\u00a0missing between the\u00a0thumb\u00a0and fingers.\u00a0\u00a0<\/p>\n

DW Fact check also<\/span>\u00a0looked through\u00a0older images provided by\u00a0SalamPix\u00a0and,\u00a0like\u00a0the\u00a0S\u00fcddeutsche\u00a0Zeitung<\/a><\/span><\/span>,\u00a0<\/span>one of Germany’s largest\u00a0<\/span>daily newspapers,\u00a0found\u00a0irregularities in images published.\u00a0\u00a0<\/p>\n

One image DW analyzed is the one at the top of the article<\/span>, allegedly showing\u00a0Iranian protesters clashing with security forces during a\u00a0protest against\u00a0the Iranian regime in Mahabad city in November 2022.\u00a0<\/p>\n

The image shows\u00a0<\/span>typical AI errors of that time: the hands,\u00a0especially the fingers, of several people in the image are deformed or look like they are made of wood. The windows of the building on the left\u00a0don’t\u00a0align at\u00a0the same angle. And the\u00a0face of the person at the far right of the image is distorted.<\/p>\n

As AI tools improve, fabricated visuals are becoming\u00a0harder to distinguish from real ones. This means that both everyday users and professional journalists are at risk of being misled \u2014 as this case\u00a0<\/span>demonstrates.\u00a0<\/p>\n

Media organizations, including DW, are investing heavily in training staff to detect and debunk AI manipulation.\u00a0<\/span>DW’s Fact check<\/a>\u00a0team also produces\u00a0<\/span>media literacy content<\/a><\/span><\/span>\u00a0to help audiences understand how to recognize misleading images and videos.<\/span>\u00a0<\/p>\n

Edited by: Thomas Sparrow, Joscha Weber, Cristina Burack<\/em><\/p>\n

[analyse_source url=”https:\/\/www.dw.com\/en\/fact-check-how-doctored-images-from-iran-misled-media-outlets\/a-76396118″]<\/p>\n","protected":false},"excerpt":{"rendered":"

[analyse_image type=”featured” src=”https:\/\/static.dw.com\/image\/76387461_6.jpg”] https:\/\/p.dw.com\/p\/5AY7S After journalists suspected that several images provided by the news agency SalamPix were AI-generated, media houses looked through their content and found many examplesImage: Salampix Manipulated or recycled photos and videos\u00a0frequently\u00a0circulate during wars and crises \u2014 sometimes to deceive, sometimes simply to generate clicks. During the ongoing\u00a0US\u2011Israel war with Iran, the […]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[2],"tags":[226,74],"class_list":["post-1834208","post","type-post","status-publish","format-standard","hentry","category-politics","tag-crawlmanager","tag-dw-com"],"_links":{"self":[{"href":"https:\/\/analyse.optim.biz\/index.php?rest_route=\/wp\/v2\/posts\/1834208","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/analyse.optim.biz\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/analyse.optim.biz\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/analyse.optim.biz\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/analyse.optim.biz\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=1834208"}],"version-history":[{"count":0,"href":"https:\/\/analyse.optim.biz\/index.php?rest_route=\/wp\/v2\/posts\/1834208\/revisions"}],"wp:attachment":[{"href":"https:\/\/analyse.optim.biz\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=1834208"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/analyse.optim.biz\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=1834208"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/analyse.optim.biz\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=1834208"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}