In a randomized test, banning AI from peer review made almost no difference to scores, and many reviewers broke their rules anyway.
Can rules against undisclosed LLM use change reviewer behavior without enforcement?
This explores whether a conference policy against using LLMs to write peer reviews changes what reviewers do, and what the reviews say, if nobody can reliably check who followed it.
This explores whether a written rule against using LLMs in peer review changes reviewer behavior when enforcement is weak or absent. The most direct evidence suggests it mostly doesn't. ICML 2026 ran a randomized experiment. Some reviewers were told not to use LLMs at all, and others were allowed limited use. The two groups ended up with almost the same paper scores, accept/reject decisions and reviewer confidence. Substantial fractions of reviewers in both groups also broke whichever rule they had been given Does banning LLM use in peer review change review outcomes?. So the rule neither stopped the behavior nor visibly changed the results.
Enforcement is hard because detection is weak. ICML planted hidden instructions in submitted PDFs that work like watermarks: if a reviewer pasted the paper into an LLM, the output would carry a telltale trace. This flagged about 795 reviews, roughly 1%, and led to 497 desk rejections. The chairs admit it mostly catches careless use and misses anyone who removed or rewrote the trace How many peer reviewers secretly used LLMs despite the ban?. Human judgment doesn't fill the gap either. Readers with machine learning expertise could not reliably tell LLM-written research abstracts from human ones, and they tended to assume a human was involved Can readers tell LLM abstracts from human ones?. ICLR 2026 took a practical approach. It treated imperfect detector flags as one input for human area chairs rather than an automatic penalty. It saved hard penalties for something that can be verified: fabricated references. Papers with confirmed hallucinated citations were desk-rejected How can conferences detect and handle LLM misuse in peer review?. The approach is to enforce against the checkable symptom, not the hidden behavior.
The lateral surprise is that a non-punitive approach changed behavior. At ICLR 2025, reviewers got optional feedback on their drafts from Claude-based agents. Over a quarter of them revised their reviews, and blinded raters judged the revisions more specific and clearer Can LLM feedback help peer reviewers improve their own reviews?. A ban asks reviewers to stay away from a tool nobody can monitor. The feedback trial put the tool where reviewers could see it and steered what they did with it. Only the second produced a measurable shift.
There is also a reason to care more about what LLM-assisted reviews do than about whether the rule was followed. LLM evaluators can be swayed by fake references and polished formatting regardless of content Can LLM judges be fooled by fake credentials and formatting?. In simulations, LLM reviewers inflated scores for LLM-written papers and marked down human papers that contained critical statements Do LLM reviewers favor papers written by other LLMs?. Yet a study of more than 125,000 real reviews found that the apparent favoritism disappears once paper quality is controlled for. LLM-assisted reviewers are just more lenient toward weaker papers, and LLM-written papers tend to be weaker Do LLM reviewers actually favor LLM-written papers?. The corpus doesn't contain a study that isolates a rule with zero enforcement. But taken together, the evidence suggests the useful lever is not whether a rule exists. It is whether a conference can verify the specific harms it worries about, or shape how the tool gets used.
Sources 8 notes
A randomized experiment at ICML 2026 found that prohibiting LLM use versus allowing limited use barely changed paper scores, decisions, or reviewer confidence. Meanwhile, substantial fractions of reviewers broke whichever rule they were given.
Hidden-instruction watermarks planted in PDFs flagged about 1% of reviews under ICML's no-LLM rule, leading to 497 desk rejections. The chairs acknowledge the method catches mainly careless uses and misses reviewers who removed or rewrote the watermark.
Readers with ML expertise struggle to identify LLM-generated content reliably, tending to assume human involvement across all abstract types. However, LLM-edited abstracts received highest clarity ratings and were preferred 55% of the time when authorship was disclosed.
Program chairs used imperfect detectors as one input for area chairs rather than automated filters, but desk-rejected papers with confirmed fabricated references as a tractable enforcement point. Multiple human review steps mitigated false positives.
A randomized trial at ICLR 2025 found that optional, gated feedback from Claude-based agents led over a quarter of reviewers to update their reviews, incorporating suggestions that blinded raters judged as more informative and clear.
Show all 8 sources
Research identified four evaluation biases in LLM judges, with authority and beauty biases being semantics-agnostic and trivially exploitable through fake references and formatting—zero-shot attacks requiring no model access or optimization.
Simulated LLM reviewers gave higher scores to LLM-written papers and downrated human papers containing critical statements, while human annotators showed no such bias. The bias traces to preference for LLM writing style and aversion to critical framing.
Across 125,000+ reviews, the apparent favoritism of LLM-assisted reviewers toward LLM papers disappears once paper quality is held constant. LLM papers cluster among weaker submissions, creating a spurious interaction driven by LLM reviewers' general leniency toward lower-quality work.
Papers this line draws on 8
The research behind the notes this line reads — ranked by how closely each paper relates.
- LLM-REVal: Can We Trust LLM Reviewers Yet?
- Do LLMs Favor LLMs? Quantifying Interaction Effects in Peer Review
- Stop Automating Peer Review Without Rigorous Evaluation
- Use and Effects of LLMs in Peer Review: A Randomized Experiment and Survey at ICML 2026
- Position: The AI Conference Peer Review Crisis Demands Author Feedback and Reviewer Rewards
- LLM or Human? Perceptions of Trust and Information Quality in Research Summaries
- Can LLM feedback enhance review quality? A randomized study of 20K reviews at ICLR 2025
- On Violations of LLM Review Policies