OpenAI's GPT-5 reveal hits a snag
OpenAI's big reveal for GPT-5 hit a snag when viewers noticed some pretty obvious errors in the charts shown during the livestream.
For example, a graph comparing coding deception rates made o3 look worse than it was—its bar was way bigger than GPT-5's, even though their scores were almost the same.
OpenAI quickly admitted to the mix-up
OpenAI quickly admitted to the mix-up. CEO Sam Altman called it a "mega chart screwup," and a team member apologized on X for what they called an "unintentional chart crime."
The incident got people talking about how tech companies need to be extra careful with data visuals at launches.
Another chart also downplayed differences between models
Another chart also downplayed differences between models, raising more eyebrows about how numbers get presented.
With Altman leading fast-paced AI releases, some experts are asking if enough caution comes with all this innovation—or if mistakes like these will keep happening.