On Friday, marketing professionals pointed out that while ChatGPT showed safety warnings for links to Republican fundraising sites, it didn’t do the same for Democratic ones. OpenAI has come under fire for what they describe as a “technical malfunction,” a topic explored in the first chapter of a new book.
A report indicated that OpenAI acknowledged a “technical error” in ChatGPT, which flagged links to Republican sites but not to Democratic ones. This issue became apparent when users found that links to WinRed, the GOP’s donation platform, were labeled as potentially unsafe, whereas links to ActBlue, the Democratic platform, flew under the radar.
Digital marketer Mike Morrison was the first to notice this discrepancy. He shared his observations on social media, detailing how ChatGPT provided a link to WinRed with a cautionary message: “Please ensure this link is secure.”
This warning indicated that the link was unverified and could contain user data that might be shared with third parties, urging users to double-check their trust in the link. However, when he looked at a link leading to an ActBlue store, there was no such warning.
“It’s strange that ChatGPT warns about [WinRed] links being potentially unsafe,” Morrison commented online, noting that the ActBlue links were labeled as trustworthy.
OpenAI quickly addressed the issue. A spokesperson stated that this situation was unexpected and they are currently investigating. Kate Waters from OpenAI explained the company’s approach to the problem.
“As soon as I came across the post, I reached out and took a closer look,” Waters noted. She emphasized that this has nothing to do with political bias but rather a technical matter.
Waters elaborated that the issue was linked to how the system processed links from certain sites. “The model generated links for both WinRed and a single instance of ActBlue that were not indexed yet, and our safety measures flagged them as AI-generated,” she clarified, assuring that they are working to rectify this permanently.
The incident triggered significant backlash among Republican officials and supporters online. WinRed CEO Ryan Lyk criticized the situation harshly, referring to it as “election interference.”
A new book discusses the left-leaning bias prevalent in Silicon Valley and how it affects AI platforms like ChatGPT. It offers insights into AI bias and suggests ways conservatives can address this issue, aiming for a balance in how fundraising links are flagged.
Sen. Marsha Blackburn described the book as essential reading, particularly noting Hall’s understanding of the conservative stance against big tech. Observers in the field like Michael Shellenberger also labeled it as important for initiating discussions on defeating authoritarian agendas linked to big tech.
read more here.
Lucas Nolan covers free speech and online censorship for specific platforms.
