Didn’t expect OpenAI to actually publish a paper, and overnight the entire internet started roasting OpenAI—all because of mathematician Tristan Buckmaster’s recent essay. This drama is now known across the web. A lot of people are worried now: the research you create using ChatGPT might be learned, read, and then plagiarized into OpenAI’s own results. According to Buckmaster’s main argument, they’ve been putting unpublished research drafts into their paid Codex sessions for a long time. OpenAI only heavily invested in the same niche technical approach after hearing rumors about their progress. OpenAI hasn’t denied it either, saying that data generated while using their products—after removing identifying information—might be used to train and improve their models, which indirectly benefits the model. As a result, OpenAI’s comment section is getting absolutely flooded.