Boyi Wei

@boyiwei

PhD student @Princeton @PrincetonCITP.

Princeton, NJ

boyiwei.com

Replies Reposts Media

Pinned
Boyi Wei
@boyiwei
Jun 3, 2025
Are static evaluations enough to reflect the risks of offensive cybersecurity agents? 🤔 We show that the answer is no! 😯 Even with minimal compute, adversaries can significantly boost offensive cybersecurity performance -- without any external assistance! 🧵👇[1/n]
GIF
1
Boyi Wei
@boyiwei
Apr 24
Great work! I like the idea of learning an option-style controller that decides when to keep or switch expert sets instead of switching experts at nearly every token. Check this out!
Zeyu Shen
@Zeyu_Shen_yo
Apr 24
New paper! 🧵 Modern MoE LLMs switch their active experts at >94% of token positions. This makes memory offloading more expensive We show you can cheaply convert pretrained MoEs (like gpt-oss-20b) into temporally extended ones, dropping switch rates from >50% to <5% while
Boyi Wei
@boyiwei
Nov 28, 2025
I'll be at #NeurIPS2025 from Dec 2 to Dec 7! Happy to catch up and meet new friends, especially those who are interested in Agents (self-improvement, scientific discovery) and AI alignment! I will also present two papers:
1
Boyi Wei
@boyiwei
Nov 21, 2025
For Bio-foundation models, data filtering cannot completely prevent the model from being misused in some cases. We show that, even for the models trained with data filtering, we can still be able to recover harmful capabilities via probing or fine-tuning. Check this out!
Udari Madhushani Sehwag
@UdariMadhu
Nov 21, 2025
Our new research from @scale_AI reveals that harmful biological knowledge can persist inside bio-foundation models even after filtering. We introduce BioRiskEval, the first comprehensive framework built to assess dual-use risk in these models using a realistic adversarial threat
Boyi Wei
@boyiwei
Oct 7, 2025
"Misevolve" can happen unintentionally when agents improve themselves, leading to undesired (mostly harmful) outcomes. As people focus more on self-improving agents, alignment strategies should also be adaptive to avoid emergent risks. Check this out if you are interested!
Qihan Ren
@jsonren00
Oct 7, 2025
[1/8] New risk in self-evolving agents: "Misevolution"—when self-evolution unintentionally deviates and causes harm. We found this in various evolutionary paths (model, memory, tool, workflow), even with SOTA LLMs. E.g. a coding agent's ASR surged from 0.6% to 20.6% after memory
1