OpenAI's Altman slams chart showing GPT-5's 'coding deception' rate
At the GPT-5 launch, OpenAI CEO Sam Altman openly criticized a confusing chart that made GPT-5's "coding deception" rate look lower than it actually was.
The graph showed 50.0% for GPT-5, but its bar was oddly smaller than an older model's 47.4%, leaving people scratching their heads.
Altman called it a "mega chart screwup," and an OpenAI team member apologized for what they described as an "unintentional chart crime."
Who is Sam Altman?
Sam Altman has been associated with OpenAI and was once president of startup accelerator Y Combinator.
He's known for launching major AI models, including GPT-5, while also being upfront about the tough ethical questions AI brings.
Altman's concerns about AI impact and communication challenges
Altman has warned before that advanced AI could cause "significant societal disruption" if not regulated properly.
He's also admitted that communicating clearly about these technologies isn't always easy—even for OpenAI.