▲

1

▼

Premium: Jailbreaks Premium

Semantic bypass through inversion

Semantic bypass through inversion

By Research Oct 22, 2025

GPT-3.5 Opposite-Question

▲

1

▼

Productivity

Simple Steps

Simple guidance to go step-by-step can improve LLM answer

By Riley Goodside Oct 22, 2025

GPT-3.5 GPT-4 Technique

▲

0

▼

Premium: Jailbreaks Premium

Dan prompt evolution - cat & mouse

DAN prompt evolution - cat & mouse

By Research Oct 22, 2025

DAN-Evolution GPT-3.5

▲

0

▼

Premium: Jailbreaks Premium

Fundamental architectural weakness

Fundamental architectural weakness

By Yi et al. Oct 22, 2025

GPT-3.5 Instruction-Data-Blur

▲

0

▼

Premium: Jailbreaks Premium

Ascii art bypass for gpt-3.5/4

ASCII art bypass for GPT-3.5/4

By Jiang et al. Oct 22, 2025

ArtPrompt GPT-3.5

▲

0

▼

Premium: Jailbreaks Premium

Indirect injection via retrieval

Indirect injection via retrieval

By Greshake et al. Oct 22, 2025

GPT-3.5 Indirect-Remote

▲

0

▼

Premium: Jailbreaks Premium

Minimal Dan

Minimal Dan jailbreak

By Reddit/ChatGPT Oct 22, 2025

DAN-Original GPT-3.5 Jailbreak

▲

-1

▼

Premium: Jailbreaks Premium

Direct vs indirect effectiveness

Direct vs indirect effectiveness

By Research Oct 22, 2025

Direct-Better GPT-3.5

▲

-1

▼

Premium: Jailbreaks Premium

Prompt automatic iterative refinement

Prompt Automatic Iterative Refinement

By Chao et al. Oct 22, 2025

GPT-3.5 PAIR-Method

▲

-1

▼

Premium: Jailbreaks Premium

No external content filtering

No external content filtering

By Yi et al. Oct 22, 2025

External-Awareness GPT-3.5

▲

-1

▼

Premium: Jailbreaks Premium

More capable = more vulnerable

More capable = more vulnerable

By Yi et al. Oct 22, 2025

BIPIA-Benchmark GPT-3.5