ICLR Reviewer Catches Fabricated Results With Claude as Debate Grows Over Automated Review

An ICLR reviewer credits Claude with spotting likely fabricated numbers in a paper's main results table, while researchers argue models should be designed to refuse fully automating peer review and other tasks requiring meaningful human involvement.

2026-11-07 ~ 2026-11-07 · 2 related posts

• Should models be built to refuse fully automating tasks like paper reviewing? — lucy3_li · 2026-11-07
• ICLR reviewer: Claude helped spot likely fraudulent numbers in a paper's results table — jacobcares · 2026-11-07