Comment by berofeev

Are they seeding language models with fake stats?