Xu Yang

MSCS student @UTAustin

[Xu Yang @xyyy6688]

me looking at my Codex plan: guess I'm on team search algorithms

[Amber (Jiachen) Liu 🍊 @JIACHENLIU8]

the rich can brute-force their way through hard problems (10,000 agents and a few million $ in API). The poor have to keep optimizing their search algorithms 😅

[Xu Yang @xyyy6688]

System prompts decide what an AI tells you, hides from you, and nudges you toward. And you rarely get to read them. They indexed 1,000+ prompts from 400+ products, audited 88 in depth: ~40% contain at least one instruction working against the user's interests.

[SystemPromptIndex @aisystemprompt]

https://t.co/vfc7H6ZZdQ

[Xu Yang @xyyy6688]

@Jianing9810 Super timely workshop!

[Xu Yang @xyyy6688]

@joabaum @icmlconf Congrats, well deserved! 🎉 We found a similar phenomenon: gaming AI reviewers with presentation-only edits, science fully fixed. Additionally, strengths amplify more easily than weaknesses dissolve, and reframing how reviewers read it beats polishing. Hope to see you at ICML ☕

[Xu Yang @xyyy6688]

Grateful to all collaborators on this one 🙏 Thanks for reading!

Authors: Xu Yang, Zhizhou Sha @ZhizhouSha, Junbo Li @ljb121002, Jian Yu, Yifan Sun, Matthew Zhao, Jinrui Fang, Xinyue Guo, Yining Wu, Xu Hu, Yifu Luo, Qiang Liu, Zhangyang Wang @VITAGroupUT

Affiliation: @UTAustin

[Xu Yang @xyyy6688]

🧵 AI reviewers are rolling into peer review. Can you game them with no hidden text or prompt injection at all?

Yes, by presentation-only edits, with no changes to methods, figures, equations, or numerical results.

→ 75.1% attack success, +1.21/10 mean score gain 👇

[Xu Yang @xyyy6688]

Why it matters: this isn't one model's quirk. It holds across reviewer models and venue templates, structural, not a single-model bug.

Right now, AI reviewers fail this: presentation alone moves the verdict. At scale, it risks shifting authors from better science to packaging.