OpenAI CEO Sam Altman has admitted he feels “scared” after testing the upcoming GPT-5 model, raising concerns over the rapid progress of artificial intelligence.
Sam Altman compared the moment to the Manhattan Project, underscoring the scale and unpredictability of what’s being built.
During the conversation, Altman said the new model “feels very fast” and admitted that while testing GPT-5.
He reflected on moments in scientific history when creators looked at their own inventions and were unsure if they had gone too far.
“Maybe it’s great, maybe it’s bad—but what have we done?” he said, echoing the unease that accompanied the development of nuclear technology in the 1940s.
Altman also acknowledged the absence of effective oversight in AI development, saying it feels like “there are no adults in the room.”
The rapid advancement of AI, he suggested, is outpacing regulatory frameworks and societal understanding.
While he did not share technical specifics about GPT-5, his choice of comparisons suggests serious ethical and safety concerns.
While details are limited, GPT-5 is expected to improve significantly over GPT-4, offering faster response times, better multi-step reasoning, and a longer memory window
Some observers see Altman’s dramatic remarks as strategic—meant to generate anticipation and concern ahead of GPT-5’s expected August release.