AI-Generated Images Fuel Misinformation and Hateful Rhetoric, Resulting in Bomb Threats in Ohio
Springfield, Ohio, has become an unexpected focal point of controversy, as AI-generated images have been used to propagate a false and inflammatory narrative regarding Haitian immigrants. Authorities, led by Ohio Governor Mike DeWine, have responded to over 33 bomb threats as of 16 September, stemming from the spread of misinformation.
The misinformation centres around unsubstantiated claims that Haitian immigrants in Springfield are eating pets. Despite the lack of credible evidence, this narrative has been amplified through social media, particularly via platforms such as Facebook, Instagram, and X. The use of generative artificial intelligence to create these misleading images has played a significant role in the spread of this false information.
One prominent figure pushing this narrative is Ohio Senator JD Vance, the Republican nominee for vice president, who reiterated the unfounded claims on X. Former President Donald Trump also mentioned the narrative during a presidential debate on 10 September, which was viewed by 67 million people. AI-generated images depicting Trump protecting animals have since been widely circulated, despite their clear fantastical nature.
One notable post from the House Judiciary GOP on 9 September featured an image of Trump cradling a duck and a cat with the caption, "Protect our ducks and kittens in Ohio!" These AI-invented images extend to portray Trump holding kittens and running from Black people or skeletons, all of which have been shared on platforms like Truth Social.
An interview with CNN’s Dana Bash saw Vance defend these images, stating, “If I have to create stories so that the American media actually pays attention to the suffering of the American people, then that’s what I’m going to do.” Social media users have similarly engaged with these images, often ignoring their authenticity and focusing on the false narrative instead.
Experts have weighed in on the issue, warning that AI-generated content can have real-world implications. Callum Hood, head of research at the Centre for Countering Digital Hate, noted that, regardless of how implausible these images appear, they still serve to promote false and harmful narratives.
“The images contribute to hateful rhetoric that may spark real-world harm,” Hood told PolitiFact. Despite some images looking obviously fake, they still reinforce negative sentiments and misinformation when combined with inflammatory messaging. For instance, one Facebook video showing cats in camouflage, captioned “Navy SEAL Kittens,” linked to the false pet-eating narrative, garnered significant engagement.
Jonathan Nagler, a politics professor from New York University, and Whitney Phillips, an assistant professor at the University of Oregon, both affirmed that while AI-generated images alone are not the sole problem, the combination of these images with hateful rhetoric increases engagement and the spread of harmful narratives. Darren Linvill, co-director of the Media Forensics Hub at Clemson University, also suggested that the emotional appeal of such imagery can heighten engagement.
Research by the University of Waterloo and Carleton University indicates that people still struggle to differentiate between real and AI-generated images, showing an average accuracy rate of just 61% amongst study participants.
The rapid development and accessibility of generative AI tools have made it easier to create and spread misinformation at an unprecedented scale. This ease of use and wide dissemination are what set these tools apart from traditional methods like Photoshop. Jevin West, co-founder of the Centre for an Informed Public at the University of Washington, stressed that the realistic nature of AI creations makes them more convincing and potentially more dangerous.
While it remains challenging to determine if the AI-generated images directly caused the bomb threats, Phillips and Hood both suggest that the overall anti-immigrant rhetoric and the use of AI to amplify such narratives have significant impacts. Governor DeWine confirmed that none of the bomb threats were validated, but they did cause substantial disruptions to public services in Springfield.
As social media platforms grapple with the proliferation of AI-generated content, measures such as labelling AI-generated images have been implemented. However, these measures fall short of solving the broader issue. Experts like Hood argue for complementary measures such as content moderation to tackle misinformation effectively. Additionally, there is an ongoing debate on how platforms should handle AI-generated content that promotes false narratives.
The situation in Springfield underlines the broader challenge posed by the intersection of AI technology and social media. As experts continue to explore solutions, the need for a balance between freedom of expression and the mitigation of harmful content remains a pressing concern.
Source: Noah Wire Services