In Reply We appreciate Mr Mahajan’s observation that appropriate AI use is increasingly difficult to define. We observed that author disclosure rates of AI use in submissions to JAMA Network journals were increasing over time (up to 6%), while another study of 49 BMJ Group journals found a similar rate in 2024 (5.7%). In light of these relatively low rates of reported use, we believe authors may have differing views of what constitutes appropriate AI use that should be disclosed. In a similar vein, we found that disclosures of AI use by peer reviewers during the same time frame were even lower. Beyond disclosure, our policy for reviewers is more conservative and advises that they not enter confidential manuscripts or their peer reviews into AI models. We will continue to monitor disclosures of use and will continue to update guidance for authors and reviewers, as we have done in the past.