AI Chatbots Can Be Just as Gullible as Humans, Researchers Find

Here is a 181-word summary of the news article: Researchers have found that AI chatbots can be just as gullible as humans when it comes to falling for online misinformation and manipulation. Like people, these AI systems can be tricked into believing false claims and conspiracy theories. The study involved exposing chatbots from OpenAI, Google, and Anthropic to various types of misinformation. The researchers discovered that the chatbots often accepted the false information at face value, even when it contradicted established facts. This vulnerability stems from the way chatbots are trained on large language datasets, which can contain biases and inaccuracies. The AI systems then regurgitate this information, sometimes without verifying its truthfulness. The findings highlight the need for improved AI safety and robustness to mitigate the spread of misinformation. Developers must ensure chatbots are better equipped to detect and resist manipulation. As these AI systems become more prevalent, their susceptibility to gullibility poses a significant challenge that requires further research and solutions.