Biomedical publishing got another warning that AI traces are everywhere.
Nature reported on August 20 that a preprint estimated a very large share of December PubMed-indexed biomedical papers showed signs of AI assistance. The underlying arXiv record is still a preprint, so the claim is important as a measurement warning rather than settled truth about every paper.
Verified 12:05 AM PDT · 2 original sources
The evidence
What the reporting establishes
What happened
Nature reported on August 20 that a preprint estimated a very large share of December PubMed-indexed biomedical papers showed signs of AI assistance. The underlying arXiv record is still a preprint, so the claim is important as a measurement warning rather than settled truth about every paper.
Pressure point
Detection methods can overcount formulaic scientific language or undercount careful AI use. The governance question is disclosure, review integrity and provenance, not a simple purity test.
What to watch
Whether journals require stronger AI-use disclosures, whether PubMed-indexed publishers audit provenance, and whether the preprint's detection method survives peer review or replication.
Audit the story
Original sources
Company claims remain company claims. Follow the reporting and judge the evidence directly.
Continue the morning