For the past year, I’ve been living and studying in Dublin, Ireland. In my efforts to acclimate, I’ve gone out of my way to consume Irish news from Irish sources. In May of this year, I came across this story:
In short, the Irish Times unknowingly published an article written by generative AI. The article, which appeared to be written by an Ecuadorian health worker living in Dublin, criticized Irish women for using fake tan to “mock” individuals with naturally dark skin tones. The article sparked debate on social media.
In their investigation into the hoax, the Guardian reported that 80% of the article written by GPT-4. Perhaps more concerning is the fact that the author of the original article – Adriana Acosta-Cortez – also appears to be AI generated. Investigation revealed that the article’s actual author created a Twitter account using the fake name, then used Dalle-E2 to create a profile picture.
Discussion about the use of generative AI to mislead people usually centers around social media, but I think it’s also worth discussing how it will affect journalism. The Irish Times is a fairly credible and professional news outlet, and yet an individual was able to convince the outlet to publish an article written by AI, while posing as someone else, using an AI-generated profile picture attached to a fake Twitter account. AI appears able to give people the ability not just to create misleading content, but also the ability to create a reasonably convincing false identity to publish it under.
Corrosive AI 