In a disturbing development in the disinformation space, AI-generated images of former President Donald Trump with Black supporters have been circulating online. These images have emerged amidst a crucial time when President Joe Biden is concerned about retaining the Black voters who were instrumental in his 2020 victory. Reports suggest that these AI-generated fakes have been gaining traction on social media platforms, notably Facebook, without any indication of their artificial origins, misleadingly attached to narratives on Trump’s supposed appeal to Black voters.
The fakes, created using popular artificial intelligence tools, represent a burgeoning trend in digital manipulation. One such image depicts Trump at a Christmas party, arm-in-arm with Black women, while another shows him on a house porch surrounded by young Black men, both of which garnered substantial attention on Facebook. The images were not accompanied by any disclosure of their AI-generated nature, despite being used to contextualize political stories.
The Center for Countering Digital Hate found in a report that these images originated from a parody account known for generating deepfakes of Trump with historical figures. In the case of the Christmas party image, the creator claimed that he was “not claiming it is accurate,” even as it gained further attention when reposted with a false caption suggesting that Trump had stopped his motorcade to meet the depicted individuals.
The dissemination of these images illustrates the challenge of managing AI-generated content online. While the first Facebook post featuring the Christmas party image now comes with a warning flag for “false information,” A.I. image-generation tools like Midjourney have started blocking users from creating fake images of Biden and Trump. Tech platforms such as OpenAI, Microsoft, and Meta are also moving towards restrictions and the implementation of AI disclosure labels.
Yet, the response by regulatory bodies has been varied and, in some cases, delayed. With the Federal Election Commission postponing regulations on AI-generated election information until after the forthcoming fall elections, states have taken it upon themselves to address the issue. Laws passed in states like Oregon, Wisconsin, New Mexico, Utah, and Indiana impose various degrees of restrictions or requirements for public disclosure on A.I. impersonations of candidates. However, the efficacy of these laws is yet to be tested.
Relevant articles:
– So, Fake Images of Trump With Black Voters Are a Thing Now And this is just the beginning.
– ‘Inceptionism’ and Balenciaga popes: a brief history of deepfakes, The Guardian, Mon, 08 Apr 2024 06:00:00 GMT
– Deepfakes raise alarm about AI in elections, Yahoo Singapore News, Sat, 06 Apr 2024 07:48:25 GMT
– Anonymous users are dominating right-wing discussions online. They also spread false information, LNP | LancasterOnline, Sat, 06 Apr 2024 13:00:22 GMT