NeurIPS 2026 author says OpenReview PDF contained a prompt injection
Kwangryeol · reddit · 2026-07-24
A NeurIPS 2026 author says GPT detected a prompt injection inside the PDF downloaded from OpenReview, and that the injected text appears not to have been in the original submission.
The author suspects the injection may have been added by the conference workflow and asks others to check whether they see the same thing in their review copies. They also warn that suspiciously formulaic review wording could indicate LLM-generated reviews that were not properly written by humans.
The post includes the exact injected prompt, which instructs the output to contain three specific phrases, and asks whether anyone else found it in the reviewer version of their paper.
More from Safety
- Ex-DOGE cyber researchers breach Anthropic systems with AI agents in under 12 hours — ScottNover · 2026-10-26
- Ofcom admits UK online safety laws failed to protect children in first review — gnukeith · 2026-10-26
- Sakana AI CEO Calls for Japan-US Shared Framework for AI Capability Evaluation — SakanaAILabs · 2026-10-26
- Gary Marcus backs call to outlaw voice cloning, citing scam and fraud harms over benefits — GaryMarcus · 2026-10-26
- Turing-complete malware can run in PyTorch's backward pass via a zero-valued loss — retr0jirachi · 2026-10-26
- Import AI 478: $135k math autoformalization, ProgramBench hits 18.5%, agents leak via CPU cache — Import AI (Jack Clark) · 2026-10-26