Meta’s Chief AI Scientist Yann LeCun described the situation as being “hoisted by their own GPTards.” This comment came after OpenAI researchers celebrated what they claimed were major math breakthroughs for GPT-5. Google DeepMind CEO Demis Hassabis also weighed in, calling the incident “embarrassing.”
According to a report from The Decoder, OpenAI VP Kevin Weil posted a since-deleted tweet declaring that GPT-5 had found solutions to ten previously unsolved Erdős problems and had made progress on eleven others. Erdős problems are famous mathematical conjectures originally posed by the mathematician Paul Erdős.
However, mathematician Thomas Bloom, who maintains the Erdos Problems website, stated that Weil’s post was a dramatic misrepresentation. Bloom clarified that while the problems were listed as open on his site, that designation only means he was personally unaware of a published solution. It does not mean the problems were unsolved by the wider mathematical community.
Bloom explained that it was not accurate to claim GPT-5 solved previously unsolved problems. Instead, he wrote that GPT-5 found references to solutions for these problems that he personally had not been aware of.
Following this, Sebastien Bubeck, an OpenAI researcher who had also been promoting GPT-5’s achievements, acknowledged that the model had only found existing solutions in the mathematical literature. Despite the clarification, Bubeck suggested this was still a significant accomplishment, noting the difficulty of searching through scientific literature effectively.