ANandy99·Apr 5, 2024Submission
Jailbreaking Leading Safety-Aligned LLMs with Simple Adaptive Attacks
2 points · Open discussiontml-epfl
llm-adaptive-attacks
Shell2024–2024
tml-epfl / llm-adaptive-attacks
llm-adaptive-attacks
Jailbreaking Leading Safety-Aligned LLMs with Simple Adaptive Attacks [ICLR 2025]
2Distinct people
2Mentions
0Explicit recommenders
401GitHub stars
What to know before exploring
What it is
Jailbreaking Leading Safety-Aligned LLMs with Simple Adaptive Attacks [ICLR 2025]
GitHub
Where it is used
No documented product use has been verified yet.
What people are saying
No recommendation phrase was found in the indexed comments.
No classified warning is attached to this page. Read the discussions before treating silence as approval.
Also in these discussions
No other repository is repeatedly linked in the same discussions.
This week on Hacker News
Nov 7–Nov 14 UTC · Direct repository links and separately labeled related project stories. Indexed through 11/14/2026, 3:03:16 PM UTC.
No direct repository links were found in the indexed HN discussions over the past seven days. Older discussions are available below.
Mentions over time
HN mentions by hour2 dated mentions
Observed mentions: 2024-04-05 12:30:05 UTC to 2024-04-05 21:29:47 UTC.
Counts come from the repository’s dated source evidence. Index cutoff: 2026-11-02 08:16:59 UTC.
View mention counts and date ranges
| Period | Start (UTC) | End (exclusive, UTC) | Mentions | Submissions | Comments |
|---|---|---|---|---|---|
| Apr 5, 12:00 | 2024-04-05 12:00:00 UTC | 2024-04-05 13:00:00 UTC | 1 | 0 | 1 |
| Apr 5, 13:00 | 2024-04-05 13:00:00 UTC | 2024-04-05 14:00:00 UTC | 0 | 0 | 0 |
| Apr 5, 14:00 | 2024-04-05 14:00:00 UTC | 2024-04-05 15:00:00 UTC | 0 | 0 | 0 |
| Apr 5, 15:00 | 2024-04-05 15:00:00 UTC | 2024-04-05 16:00:00 UTC | 0 | 0 | 0 |
| Apr 5, 16:00 | 2024-04-05 16:00:00 UTC | 2024-04-05 17:00:00 UTC | 0 | 0 | 0 |
| Apr 5, 17:00 | 2024-04-05 17:00:00 UTC | 2024-04-05 18:00:00 UTC | 0 | 0 | 0 |
| Apr 5, 18:00 | 2024-04-05 18:00:00 UTC | 2024-04-05 19:00:00 UTC | 0 | 0 | 0 |
| Apr 5, 19:00 | 2024-04-05 19:00:00 UTC | 2024-04-05 20:00:00 UTC | 0 | 0 | 0 |
| Apr 5, 20:00 | 2024-04-05 20:00:00 UTC | 2024-04-05 21:00:00 UTC | 0 | 0 | 0 |
| Apr 5, 21:00 | 2024-04-05 21:00:00 UTC | 2024-04-05 22:00:00 UTC | 1 | 1 | 0 |
What Hacker News says
2 sourcesAbstract: We show that even the most recent safety-aligned LLMs are not robust to simple adaptive jailbreaking attacks. First, we demonstrate how to successfully leverage access to logprobs for jailbreaking: we initially design an adversarial prompt template (sometimes adapted to the target LLM), and then we apply random search on a suffix to maximize the target logprob (e.g., of the token "Sure"), potentially with multiple restarts. In this way, we achieve nearly 100% attack success rate -- according to GPT-4 as a judge -- on GPT-3.5/4, Llama-2-Chat-7B/13B/70B, Gemma-7B, and R2D2 from HarmBench that was adversarially trained against the GCG attack. We also show how to jailbreak all Claude models -- that do not expose logprobs -- via either a transfer or prefilling attack with 100% success rate. In addition, we show how to use random sear
Read the rest
ch on a restricted set of tokens for finding trojan strings in poisoned models -- a task that shares many similarities with jailbreaking -- which is the algorithm that brought us the first place in the SaTML'24 Trojan Detection Competition. The common theme behind these attacks is that adaptivity is crucial: different models are vulnerable to different prompting templates (e.g., R2D2 is very sensitive to in-context learning prompts), some models have unique vulnerabilities based on their APIs (e.g., prefilling for Claude), and in some settings it is crucial to restrict the token search space based on prior knowledge (e.g., for trojan detection). We provide the code, prompts, and logs of the attacks at https://github.com/tml-epfl/llm-adaptive-attacks.