Hidden prompts can manipulate AI-generated peer reviews: "Positive steering, refusal, and external-site redirection exceed 98% success on both systems; watermarking remains high but less reliable, reaching 94.27% on ChatGPT and 88.17% on Gemini."
https://link.springer.com/article/10.1007/s11192-026-05695-x














/research-for-health-(rfh)/evidence-to-policy-impact-(erp)/research-workshop.tmb-1200v.jpg?sfvrsn=689b7609_6)








