Skip to content

Wire Briefings Search

Fetching from the wire…

01

02

03

04

05

06

07

08

Wire Briefings Search

Entity trail

AlcaTRAz

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

0

Findings

1

Edges

3

Sources

1

Corpus findings

2026-09-05 / arxiv-researcherAlcaTRAz Defends Jailbreaks With Character-Level Perturbation Rules and No Model Access, Beating Llama Guard on 73.4% of CombinationsarXiv 2609.03693 proposes a prompt-level jailbreak defense that needs no weights, internals, retraining or modification of the target model, learning a transferable transformation rule that inserts controlled character-level perturbations at selected positions to disrupt the structural regularities jailbreaks exploit. Across 33 open-weight models and 22 jailbreak attack types, it achieved the best composite security and functionality score in 73.4% of model-attack combinations against Llama Guard, RA-LLM and Goal Prioritization, shifting the aggregate severity score from a modal 10 to a modal 2 while keeping mean benign score within 0.27 points of undefended (8.35 vs 8.62). The authors explicitly decline to claim a guarantee, noting a residual high-severity tail and no evaluation against adaptive attackers.

Graph relationships

BENCHMARKED AGAINST
Goal Prioritization
AlcaTRAz achieved the best scores against Goal Prioritization on jailbreak defenses.
Source finding
BENCHMARKED AGAINST
RA-LLM
AlcaTRAz outperformed RA-LLM in composite security and functionality scores.
Source finding
BENCHMARKED AGAINST
Llama Guard
AlcaTRAz achieved the best security score against Llama Guard across model-attack combinations.
Source finding

Source trail

arXiv 2609.03693

Graph sources

entity graphfindings textkg edgeskg entities