Paper

Jailbreaking for the Average Jane: Choosing Optimal Jailbreaks via Bandit Algorithms for Automatically Enhanced Queries

Unreadunread

AI summary

arXiv:2606. 26936v1 Announce Type: new Abstract: With a profusion of jailbreaks for LLMs now widely known, a growing concern is that non-expert malicious actors ("the average Jane") could elicit actionable responses to malicious requests.

Reading status

Priority

About this paper

Authors: Prarabdh Shukla, Ritik, Suhas Rao, Arpit Agarwal, Arjun Bhagoji
Released: 2026-06-26 (1mo ago)
Topic: Related to current project
Relation to my work: Read next because Jailbreaking for the Average Jane: Choosing Optimal Jailbreaks via Bandit Algorithms for Automatically Enhanced Queries overlaps with clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)", experiment "#351 follow-up: broader-vocab position-0 sweep at T=1.0 + position-1 suffix isolation", experiment "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: line, rate, lora, model. Source: arxiv cs.CR (Cryptography and Security).
arXiv: 2606.26936

Ask Claude about this paper

Posts a comment with @claude so the runner spins up a Claude Code session scoped to Jailbreaking for the Average Jane: Choosing Optimal Jailbreaks via Bandit Algorithms for Automatically Enhanced Queries. Replies appear in the conversation below.

Comments

No comments yet.

Related records

No links yet.