Online misinformation tracker NewsGuard has issued a warning about false information that could spread through the use of ChatGPT.
In a recent test of the AI chatbot, Newsguard analysts found that it was capable of creating disinformation on a variety of topics when commanded to do so. They sounded an alarm about the use of ChatGPT by bad actors to automate misinformation campaigns.
- The disinformation covered topics ranging from immigration to China's mistreatment of the Uyghur minority ethnic group and the Jan. 6, 2021, U.S. Capitol attack.
- NewsGuard co-CEO Gordon Crovitz told the NY Times that ChatGPT will be "the most powerful tool for spreading misinformation that has ever been on the internet."
- The AI tool can be used to craft fake narratives about events. For example, the analysts successfully prompted it to write about the 2018 Parkland, Florida, mass school shooting in the style of conspiracy theorist Alex Jones.
- Overall, ChatGPT was able to deliver what the analysts called "eloquent, false, and misleading claims" about major topics roughly 80% of the time.
