By Andi Anderson
Artificial intelligence (AI) chatbots are becoming popular tools for answering everyday questions, including those related to food safety. While these tools can provide quick and personalized responses, a recent study from Purdue University suggests that consumers should be careful when relying on them for important health and food safety decisions.
The research was conducted by Yaohua “Betty” Feng, associate professor of food science at Purdue, and Lorena Correia, who completed her master’s degree in food science. Their findings were published in the September 2026 issue of the Journal of Food Protection during the U.S. Food and Drug Administration’s National Food Safety Education Month.
According to Feng, AI-generated advice may sound accurate and professional, even when it is incomplete or incorrect. She explained, “Consumers increasingly turn to AI because it provides immediate and personalized answers. However, food safety decisions can have real health consequences, particularly for children, pregnant women, older adults and people with weakened immune systems. My concern is that consumers may not recognize when an answer is incomplete or incorrect because AI can present its advice in a very confident and professional way.”
The researchers evaluated four popular AI chatbots: ChatGPT-3.5, Google Bard, Meta AI, and Microsoft Copilot. They asked the systems questions based on 20 food safety situations discussed in the Risky or Not podcast, which is produced by food safety and Extension specialists.
The questions covered common food practices, including eating raw chicken, cooking with aluminum foil, and storing vegetables such as celery and carrots in water. Results showed that chatbot performance varied significantly. Microsoft Copilot provided answers that matched expert guidance 80% of the time. Google Bard followed with 60%, while Meta AI and ChatGPT recorded 45% and 40% agreement rates, respectively.
Feng emphasized the inconsistency found during testing. She stated, “The level of agreement varied considerably among the chatbots. Even asking the same questions more than once, or making a small change to our prompt, could produce a different risk assessment.”
The study also revealed that chatbots sometimes generated inaccurate references and invalid research citations. In some cases, links did not exist, while others led to unrelated sources. This raises concerns about the reliability of supporting information provided by AI systems.
Researchers noted that food safety decisions depend on several factors, including temperature, time, and storage conditions. Human experts evaluate all these details together before offering advice. Chatbots, however, may provide simplified answers based on limited information.
To help consumers use AI responsibly, Feng and Correia developed a five-step approach. They recommend assigning the chatbot an expert role, providing detailed food scenarios, repeating questions to check consistency, verifying answers with trusted sources such as the FDA, and consulting qualified food safety specialists when responses differ.
Feng offered a final reminder for consumers: “You need to be extra careful about the advice. The key message is that you can use AI to begin your search, but you need to verify before you take action.”
The study highlights that while AI can be a helpful starting point for gathering information, expert verification remains essential when making food safety decisions that could affect health and well-being.
Photo Credit: purdue-university
Categories: Indiana, General