A study by Newsguard reveals that the 10 most popular AI chatbots provide false information in one out of three answers. Inflection AI’s Pi and Perplexity AI have the highest rates of false claims, while Anthropic’s Claude and Google’s Gemini have the lowest. The report highlights a significant increase in falsehoods from Perplexity since 2024 and notes that some chatbots cite foreign propaganda as sources. Despite claims of improved accuracy, the models continue to fail in similar areas as before. Teljes cikk (Euronews.com)