I think the dataset/benchmark "FrontierMath" is out there. Just no evidence of this "secret meeting" taking place, or the results of it. At least nothing I can find by clicking on random links in the article.