Non-review LLM-generated arXiv papers outnumber review ones nearly 6x
yanaiela · x · 2026-09-10
A study by Yanai Elazar and Maria Antoniak (EMNLP 2026 Findings) quantitatively tests arXiv's ban on position papers. Using two LLM-content detectors, they find LLM-generated content rose sharply in both review and non-review papers, with higher prevalence in reviews — but by paper count, non-review LLM-generated papers are estimated at nearly 6x the review ones. The policy could cut up to 50% of papers in Computers & Society. The authors argue the ban lacked quantitative evidence and released their analysis code.
Related event: Study finds arXiv's LLM paper restrictions lack empirical basis(4 posts)→
More from Safety
- Ex-DOGE cyber researchers breach Anthropic's systems with AI agents in under 12 hours — GarrisonLovely · 2026-10-27
- Alignment Hub launches with 120+ documented AI misbehavior incidents and a reporting template — Thom_Wolf · 2026-10-27
- Alignment Hub launches to track 37 misalignment incidents with disclosure stats — eliebakouch · 2026-10-27
- US senator slams AI industry for proposing 50-death threshold for 'catastrophic incident' — peterwildeford · 2026-10-27
- 125 poisoned samples backdoored Qwen2.5-7B — and open-weight supply chain risks debated — lxrjl · 2026-10-27
- Dev training a guard model to block prompt injection, tool and memory poisoning — gnukeith · 2026-10-27