LESSWRONG
LW

749

Jailbreaking (AIs)

This page is a stub.

Posts tagged Jailbreaking (AIs)

50Role embeddings: making authorship more salient to LLMs

Nina Panickssery, Christopher Ackerman

9mo

0

37Illusory Safety: Redteaming DeepSeek R1 and the Strongest Fine-Tunable Models of OpenAI, Anthropic, and Google

ChengCheng, Brendan Murphy, Adrià Garriga-alonso, Yashvardhan Sharma, dsbowen, smallsilo, Yawen Duan, ChrisCundy, Hannah Betts, AdamGleave, Kellin Pelrine

8mo

0

12A Poem Is All You Need: Jailbreaking ChatGPT, Meta & More

Sharat Jacob Jacob

1y

0

9Interpreting the effects of Jailbreak Prompts in LLMs

1y

0

1Jailbreaking Claude 4 and Other Frontier Language Models

4mo

0

18Detecting out of distribution text with surprisal and entropy

9mo

4

17Constitutional Classifiers: Defending against universal jailbreaks (Anthropic Blog)

8mo

1

16Intriguing Properties of gpt-oss Jailbreaks

zroe1, Jack Sanderson

2mo

0

10Using hex to get murder advice from GPT-4o

Laurence Freeman

1y

5

9Can Persuasion Break AI Safety? Exploring the Interplay Between Fine-Tuning, Attacks, and Guardrails

8mo

0

7Breaking GPT-OSS: A brief investigation

1mo

0

4Jailbreaking ChatGPT and Claude using Web API Context Injection

1y

0

3Philosophical Jailbreaks: Demo of LLM Nihilism

5mo

0

Add Posts