AI Risk Clock
Doomsday Clock2 min to midnight
→
⚖️ Neutral edit
yahoo.com2026-05-24

AI hallucinations are infiltrating expert work—and entering the permanent body of knowledge - Yahoo

SafetyResearchModels

The study reports that in 2024, one in every 458 academic papers contained at least one fake reference, a significant increase from one in every 2,828 papers in 2023.

The study's authors attribute the rise to the growing use of large language models by researchers, who may inadvertently include fabricated citations generated by the AI. These 'hallucinations' often appear as plausible but nonexistent references, evading detection during peer review. The rapid increase from 2023 to 2024 suggests the problem is accelerating alongside AI adoption in academia.

The article notes that this trend poses challenges to the integrity of scientific literature. While AI tools can assist with literature reviews and writing, the uncritical incorporation of their outputs risks polluting the scholarly record. The authors call for better verification practices and increased awareness among researchers about the limitations of generative AI.

Read this story in another voice
● REC · 2026