NewsGuard issued its one-year AI audit progress report on September 4, 2025, finding that AI models spread falsehoods in 35% of news-related responses, with the average failure rate doubling over the past year. The report ranks chatbots by their performance in handling false claims.