The presence of AI writing is not itself an indication that a scientific paper is good, bad, or fraudulent. If we want AI to help accelerate science, we won’t do that without AI work “contaminating” journals. We should be doing more to say what good AI use in science looks like.