Comment by mullingitover

This is great, but I can't help but wonder if we're going to have another post next week with a lab complaining that they were about to publish this same finding, and they had Claude proofread their paper, and whoops how'd that get into Anthropic's training data?