RESEARCH

→ View original post on X — @sassoftware

SAS Hackathon 2024 Results Announced Innovation Winners

By

@sassoftware

–

13 November 2024 18h49

Exciting news! The 2024 #SASHackathon results are in, highlighting the most innovative teams across all categories. Who will emerge as the top contenders? Tune in on Nov. 20 at 9:00am EST | 3:00pm CET | 7:30pm IST | 10:00pm SGT http://
2.sas.com/6013sHfS7

→ View original post on X — @juanmerodio

Learning Through Practice: Transform Knowledge Into Experience

By

@juanmerodio

–

13 November 2024 18h33

Porque el verdadero aprendizaje no viene solo de escuchar, sino de hacer. Al involucrarte, transformas el conocimiento en experiencia y cada paso en un avance. Únete y haz de cada enseñanza un logro que permanece.

Research Engineer Role: Jailbreak Robustness at Anthropic

By

–

13 November 2024 18h32

If you’re interested in working on jailbreak robustness with researchers like @MrinankSharma and @EthanJPerez
, we encourage you to apply for our Research Engineer/Scientist role:

Open-Sourcing AI Robustness Benchmark for Rapid Response

By

–

13 November 2024 18h32

We’re open-sourcing our benchmark so that others can build on our work. Making systems perfectly robust might not be possible. Rapid response is a more tractable option. Paper: https://
arxiv.org/abs/2411.07494
GitHub:

LLM Jailbreak Robustness and Safety Enhancements

By

–

13 November 2024 18h32

Our work suggests that jailbreak rapid response, along with other enhancements in jailbreak robustness, offers a promising pathway for making real-world LLMs safer. We expect further gains with improved jailbreak proliferation techniques.

LLM Jailbreak Proliferation and Defense Scaling Strategies

By

–

13 November 2024 18h32

The key is producing new jailbreak-like text from a known jailbreak. We “proliferate” jailbreaks by using an LLM to generate more jailbreak examples. The best defense scales dramatically with better proliferation.

Benchmark Defense Against AI Jailbreak Attacks

By

–

13 November 2024 18h32

In the paper, we develop a benchmark for these defenses. From observing just one example of a jailbreak class, our best defense—fine-tuning an input classifier—reduces jailbreak success rate by 240× on previously detected attacks, and 15× on diverse variants of those attacks.

Adaptive Jailbreak Defense: Rapid Response Research

By

–

13 November 2024 18h32

New research: Jailbreak Rapid Response. Ensuring perfect jailbreak robustness is hard. We propose an alternative: adaptive techniques that rapidly block new classes of jailbreak as they’re detected. Read our paper with @MATSprogram
: https://
arxiv.org/abs/2411.07494