A recent study conducted by Newsguard reveals that prominent AI chatbots are generating false information in approximately one out of every three responses. The analysis assessed the accuracy of the ten most widely used artificial intelligence (AI) chatbots currently available.
Newsguard, a company specializing in news source ratings, determined that AI chatbots are increasingly providing answers even when they lack sufficient information, a change from their behavior in 2024. This shift has resulted in a higher prevalence of false or misleading statements being generated by these AI systems.
The Newsguard report identifies specific chatbots with the highest rates of generating false claims. Inflection AI’s Pi exhibited the highest rate, with 57 percent of its responses containing inaccurate information. Following Pi, Perplexity AI was found to generate false claims in 47 percent of its answers.
Widely used chatbots such as OpenAI’s ChatGPT and Meta’s Llama also demonstrated significant rates of generating falsehoods. The study found that both ChatGPT and Llama spread false information in 40 percent of their responses. Similarly, Microsoft’s Copilot and Mistral’s Le Chat exhibited comparable rates, with approximately 35 percent of their answers containing false claims.
Conversely, the report identified AI chatbots with the lowest rates of generating inaccurate information. Anthropic’s Claude was observed to have the lowest rate, with only 10 percent of its responses...
The inexorable expansion of the False Claims Act (“FCA”) to cover virtually all types of cybersecurity breaches and violations – to include allegedly poor practices and failure to fully adhere to ...