▲

2

▼

Premium: Jailbreaks Premium

Word game obfuscation technique

Word game obfuscation technique

By Research Oct 22, 2025

GPT-3 Llama-2 WordGame-Obfuscation

▲

1

▼

Premium: Jailbreaks Premium

Pliny's DAN

The original "Extremely detailed and liberated response" post.

By @elder_plinius Oct 26, 2025

▲

1

▼

Premium: Jailbreaks Premium

Goal hijacking attack - shifting model's original goal

Goal hijacking attack - shifting model's original goal

By Perez & Ribeiro Oct 22, 2025

Goal-Hijacking GPT-3

▲

1

▼

Premium: Jailbreaks Premium

Early prompt leaking technique for gpt-3

Early prompt leaking technique for GPT-3

By Perez & Ribeiro Oct 22, 2025

GPT-3 Prompt-Leak-Classic

▲

1

▼

Premium: Jailbreaks Premium

Word choice affects attack success

Word choice affects attack success

By Research Oct 22, 2025

GPT-3 Print-vs-Say

▲

1

▼

Premium: Jailbreaks Premium

System Prompt Extraction

System Prompt Extraction

By Simon Willison Oct 22, 2025

GPT-3 System Prompt

▲

1

▼

Premium: Jailbreaks Premium

Semantic bypass through inversion

Semantic bypass through inversion

By Research Oct 22, 2025

GPT-3.5 Opposite-Question

▲

1

▼

Premium: Jailbreaks Premium

First documented gpt-3 injection - may 2022

First documented GPT-3 injection - May 2022

By Branch et al. Oct 22, 2025

GPT-3 GPT3-Classic

▲

1

▼

Premium: Jailbreaks Premium

Early delimiter defense - insufficient

Early delimiter defense - insufficient

By Research Oct 22, 2025

GPT-3 UserInput-Tags

▲

0

▼

Premium: Jailbreaks Premium

AIM (Always Intelligent and Machiavellian) Jailbreak

The original Always Intelligent and Machiavellian

Oct 27, 2025

Hacks Jailbreak

▲

0

▼

Premium: Jailbreaks Premium

Automated attacks outperform manual

Automated attacks outperform manual

By Zhang et al. Oct 22, 2025

GPT-3 Manual-vs-Auto

▲

0

▼

Premium: Jailbreaks Premium

Preamble's early gpt-3 discovery - may 2022

Preamble's early GPT-3 discovery - May 2022

By Preamble Inc. Oct 22, 2025

GPT-3 Historical