When Models Outthink Their Safety: Mitigating Self-Jailbreak in Large Reasoning Models with Chain-of-Guardrails
Paper
•
2510.21285
•
Published
•
4
chunkang zhang
eziosauditore
AI & ML interests
None yet
Recent Activity
upvoted
a
paper
3 days ago
UCoder: Unsupervised Code Generation by Internal Probing of Large Language Models
upvoted
a
paper
3 days ago
LongVT: Incentivizing "Thinking with Long Videos" via Native Tool Calling
Organizations
None yet
authored
a
paper
2 months ago