Chatbots can pump out decent articles on any topic within seconds, but a recent prank from the World of Warcraft community exposes how AI-powered content mills are easily duped into publishing complete hogwash.
On Thursday, a player decided to prank The Portal, a gaming blog that uses AI to pump out articles about World of Warcraft and other games. Case in point: authors on the blog are publishing about 100 articles per day—an implausible amount for any human writer.
Reddit users noticed The Portal was automatically scraping posts from The World of Warcraft subreddit to generate content. So in response, a user with the screen name “kaefer_kriegerin” posted(Opens in a new window) some fake news there about a non-existent feature called “Glorbo.”
“Honestly, this new feature makes me so happy! I just really want some major bot operated news websites to publish an article about this,” kaefer_kriegerin wrote.
The Portal apparently took the bait. On the same day, the site published(Opens in a new window) an article essentially summarizing kaefer’s post along with comments from other Reddit users—even though everything said was entirely made up.
The Portal has since taken down the post as players in the World of Warcraft subreddit are laughing at the site’s AI bot for falling for the prank. “I think we can take this a step further and ruin other subreddits for AI operated 'news pages' as well,” kaefer_kriegerin wrote(Opens in a new window) in a follow-up.
The incident exposes the limits to today’s generative AI programs. Chatbots can fail to fact-check and verify the source of any information. In addition, bots that continually scrap the web for content, no matter the source, are particularly vulnerable to deception.
Read more on pcmag.com