OpenAI’s ChatGPT-4 has been accused of blacklisting conservative outlets, including Breitbart News, according to a Twitter user known as Elephant Civics. The user claims to have discovered a list of websites that ChatGPT-4 refuses to draw information from, citing reasons such as conspiracy theories and hate speech. It seems that ChatGPT-4’s AI safety measures and filters are designed to ensure ethical, legal, and quality responses. However, if the blacklist exists, it suggests that conservative sources are excluded from shaping the chatbot’s worldview and delivering responses.
Elephant Civics stumbled upon the alleged blacklist by asking ChatGPT-4 to provide a list of credible and non-credible news sources. Through clever prompts, the user was able to coax the chatbot into referring to a Transparency Log containing the blacklisted sites. Elephant Civics, a self-professed Republican and computer science enthusiast, claims that the list includes Breitbart News for hyper-partisan and misleading information, as well as the Epoch Times for misinformation and conspiracy theories.
Critics point out that ChatGPT-4 has been prone to making up citations and facts in the past, which is known as hallucinating in the AI industry. OpenAI may argue that the alleged Transparency Log doesn’t actually exist. This is not the first time that ChatGPT has been accused of bias, with previous versions refusing to adopt the styles of Breitbart News or Fox News, displaying selective definitions, and exhibiting political preferences.
While later versions of ChatGPT have aimed to address these issues, OpenAI has not provided detailed explanations for their occurrence. A recent university study also highlighted persistent leftist bias in the popular AI chatbot. OpenAI declined to comment on these specific allegations when approached by Breitbart News.
As discussions around AI bias continue to capture attention, concerns about the extent to which technology can reflect the ideological leanings of its developers and datasets are warranted. It remains to be seen how OpenAI will respond to these claims and whether they will take steps to ensure greater transparency and fairness in the future.