Amnesty International recently published a report on human rights abuses by the Colombian police and promoted it on social media. For this purpose, stylized images were used that were generated by software using methods of artificial intelligence. Criticism then hailed and the organization responded. The images have been removed.
The background is the riots in Colombia in 2021, when a government tax reform led to massive protests to which security forces reacted violently. In addition, numerous people are said to have been kidnapped and mistreated by the police and military. Amnesty International has worked through this in a report that was referred to with pictures on Twitter and in an Instagram story.
AI images with bugs
No real images were used for this, the human rights organization had them created by artificial intelligence. On the one hand, this was pointed out by the caption (“Illustrations produced by artificial intelligence”), but it is also relatively easy to see if you look closely. So a woman was shown with a flag in police custody, but it wasn’t a Colombian flag. The colors are correct, but the three stripes are in the wrong order.
Photojournalists and media scholars then warned Amnesty International that the use of AI imagery could undermine the organization’s work and fuel conspiracy theories, such as The Guardian reported. Photographers, in particular, feel derided by the organization’s actions, which would expose them to the dangers of taking pictures in the midst of such violent protests.
deletion and apology
Amnesty International says it has used real photos in previous reports, but in this case used AI images to protect the subjects from possible state persecution. AI images are assembled by appropriate software from various human-made photos and generated into a new image based on a given theme. This in turn raises questions about copyright, because the AI does not name the original sources.
Nevertheless, Amnesty International has now removed the images from social media and apologized. The the organization’s Norwegian Twitter account once again refers to personal protection and also explains that the Colombian security forces partially interrupted the power supply in affected areas during the protests. The darkness would have made it difficult for photographers.
Amnesty International itself is concerned with the negative consequences of using artificial intelligence, including the spread of false information. The organization must be more careful in using these technologies, although in this case they cannot do any harm.