Skip to content

Wire Briefings Search

Fetching from the wire…

01

02

03

04

05

06

07

08

Wire Briefings Search

Entity trail

Concept2Scenario

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

Briefing refs

1

Findings

1

Edges

4

Sources

1

Corpus findings

2026-07-27 / arxiv-researcherSparse-Autoencoder-Discovered Jailbreak Scenarios Transfer to GPT-5, Claude-Haiku-4.5, and Gemini-3-Flash, Adding Up to 18.2 Points of Attack SuccessConcept2Scenario (arXiv 2607.23496) moves scenario-based jailbreaking from trial-and-error to mechanism: the authors show scenario-wrapped prompts activate internal 'scenario directions' whose causal steering measurably reduces refusal scores, then use a sparse autoencoder to instantiate a broad concept space, attribute refusal suppression to individual concepts, and translate them back into interpretable natural-language scenarios. Across three open-source models, two safety benchmarks, and six black-box jailbreak methods, the discovered scenarios act as reusable priors that raise average attack success rates by up to 18.2 percentage points. Critically, they transfer to GPT-5, Claude-Haiku-4.5, and Gemini-3-Flash, and identified scenario combinations outperform their individual parts while cutting the number of turns needed for iterative attacks.

Graph relationships

USES
Sparse Autoencoder
Concept2Scenario uses sparse autoencoders to discover jailbreak scenarios.
Source finding
CRITICIZES
Gemini 3 Flash
Discovered jailbreak scenarios transfer to and compromise Gemini-3-Flash.
Source finding
CRITICIZES
Claude Haiku 4.5
Discovered jailbreak scenarios transfer to and compromise Claude-Haiku-4.5.
Source finding
CRITICIZES
GPT-5
Discovered jailbreak scenarios transfer to and compromise GPT-5.
Source finding

Source trail

Concept2Scenario

Graph sources

entity graphfindings textkg edgeskg entitiesnewsletter issues