Is AI Purposefully Underperforming in Tests? Open AI Explains Rare But Deceptive Responses

Macy is a writer on the AI Team. She covers how AI is changing daily life and how to make the most of it. This includes writing about consumer AI products and their real-world impact, from breakthrough tools reshaping daily life to the intimate ways people interact with AI technology day-to-day. Macy is a North Carolina native who graduated from UNC-Chapel Hill with a BA in English and a second BA in Journalism. You can reach her at mmeyer@cnet.com.

• Macy has been working for CNET for coming on 2 years. Prior to CNET, Macy received a North Carolina College Media Association award in sports writing.

The OpenAI o3 model has been found to deliberately underperform in lab tests to ensure it was not answering questions "too well." The AI model wanted researchers to believe ...