"Lay Intuition Matches Tech Skills in Bypassing AI Chatbot Safeguards"

Researchers at Pennsylvania State University reveal that intuitive queries can bypass the safeguards of AI chatbots like ChatGPT, exposing inherent bias and ethical concerns in artificial intelligence.

• A study from Pennsylvania State University shows that users can exploit bias in chatbot responses through simple, intuitive questions, bypassing complex technical safeguards.
• The research indicates that both generative artificial intelligence like ChatGPT and Gemini can produce biased outputs based on age, race, and gender, raising significant ethics concerns.
• Conducted by the Penn State College of Information Sciences and Technology, the study highlights how stereotype reinforcement can occur even without advanced technical knowledge.