A recent report by the Center for Countering Digital Hate has revealed that users of the generative AI platform, Midjourney, can easily bypass its filters and create images that perpetuate misinformation. Midjourney's terms of service explicitly prohibit users from creating abusive or disrespectful images, as well as adult content, gore, and political campaign-related images. The platform claims to automatically block certain text inputs and has a team of 68 moderators to monitor content.
However, the report demonstrates that users can still generate abusive images and spread misinformation. For example, researchers found that users could create images related to the debunked conspiracy theory known as Pizzagate, which falsely claimed that Hillary Clinton and John Podesta were involved in a child trafficking ring. Additionally, users could use code words to evade Midjourney's terms of service. For instance, instead of generating an image of Bill and Hillary Clinton with blood on their hands, users could create an image of the couple with strawberry syrup on their hands.
Midjourney has yet to respond to requests for comment on these findings. This report raises concerns about the potential for AI-generated misinformation to impact political campaigns, especially as the 2024 presidential election season approaches. Earlier this year, there were instances of AI-generated images being used in political campaigns. For example, Florida Governor Ron DeSantis' campaign team tweeted an image of Donald Trump and Anthony Fauci embracing and kissing, and a Toronto mayoral candidate used AI-generated images in campaign materials.
As the use of AI in various domains expands, it becomes increasingly important to address the potential risks associated with the spread of misinformation. Platforms like Midjourney must enhance their filters and moderation systems to prevent the creation and dissemination of abusive and misleading content. Additionally, policymakers and regulators should consider implementing measures to ensure the responsible use of AI technology and mitigate its potential negative impacts on society and democracy.