Papers › What is in Your Safe Data? Identifying Benign Data that Breaks Safety

What is in Your Safe Data? Identifying Benign Data that Breaks Safety

1 Apr 2024arXiv:2404.01099archive 2025-07-28

Luxi He, Mengzhou Xia, Peter Henderson

Current Large Language Models (LLMs), even those tuned for safety and alignment, are susceptible to jailbreaking. Some have found that just further fine-tuning an aligned model with benign data (i.e., data without harmful content) surprisingly leads to substantial degradation in safety. We delve into the data-centric aspects of why benign fine-tuning inadvertently contributes to jailbreaking. First, we represent fine-tuning data through two lenses: representation and gradient spaces. Additionally, we propose a bi-directional anchoring method that, during the selection process, prioritizes data points that are close to harmful examples and far from benign ones. Our approach effectively identifies subsets of benign data that are more likely to degrade the model's safety after fine-tuning. Training on just 100 of these seemingly benign datapoints surprisingly leads to the fine-tuned model affirmatively responding to >70% of tested harmful requests, compared to <20% after fine-tuning on randomly selected data. We also observe that the selected data frequently appear as lists, bullet points, or math questions, indicating a systematic pattern in fine-tuning data that contributes to jailbreaking.

PaperPDFCodeCode Syntology ran

In Syntology View this paper on Syntology, its page in Syntology's graph. That page adds which of its functions are also present in other papers' code Syntology holds, with dates, and its dated run record as JSON and a README badge. It also lists the repositories linked to the paper, the abstract and the calls for agents.

Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

For agents, on Syntology's MCP service (how to connect):

Code

Syntology 14 code samples harvested from 1 repository linked to this paper: 7 ran (of which 0 constructed an object rather than computing a result; 3 with no instrument failure: 0 honoured, 1 violated, 2 with no contract checked; 4 where Syntology's instrument failed) · 7 unverified. Where Syntology's instrument failed: 3 where our contract draft was wrong; 1 where the contract check could not run; failures of Syntology's instrument, not of the code.

By repository: official repository: 14 samples from 1 repository, 7 ran (3 with no instrument failure; 4 where Syntology's instrument failed). The run record, sample by sample. “Ran” means executed on a synthesized input, not that the code is correct or reproduces the paper.

Levels written between 2026-09-24 and 2026-09-26 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

princeton-nlp/benign-data-breaks-safety officialmentioned in papermentioned on GitHubpytorchSyntology: 14 samples, 7 ran (4 instrument-failed) · 7 unverified report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

14 samples harvested: 7 ran (of which 0 constructed an object rather than computing a result; 3 with no instrument failure: 0 honoured, 1 violated, 2 with no contract checked; 4 where Syntology's instrument failed) · 7 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Syntology's record for this page last changed ; when this build read Syntology's graph is in the build record. That date is not when the samples ran.

1ran · violated contract
3ran · our draft was wrong
1ran; the contract check could not run
2ran
7unverified

Licence: 14 of the 14 samples are pointer only, meaning Syntology does not serve that copy's text. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from princeton-nlp/benign-data-breaks-safety. “Ran” means the sample executed on a synthesized input. It does not mean the output is correct, and nothing here reproduces the paper's results. A line beginning “fixture:” under a sample that ran gives the arguments Syntology's fixture called the sample with, each as name:type/rank/element kind (? where none was recorded), and the type of what the fixture's run returned, exactly as the harness recorded them; the arguments and the returned type were recorded by two separate runs of the same fixture. No output value is shown, and a returned type says nothing about whether the output was right. “Level recorded” gives the UTC date when Syntology's graph wrote the sample's verification level, not when the code executed; where the level was written before Syntology's graph kept that date, the row says “level recorded, date not recorded”. “Honoured” and “violated” refer to a contract Syntology drafted from the code itself. “Our draft was wrong” and “the contract check could not run” are failures of Syntology's instrument, not of the code: the sample ran on Syntology's synthesized input, and then the drafted contract did not describe it, or the check itself raised an error (RAISES), lacked a dependency (DEP_MISSING) or ran out of time (TIMEOUT); hover the label for which. A metamorphic tier, where shown, is evidence of the output's shape, not that the sample implements the method.

Each sample ends with its code_sha256. A sample's code_sha256 is Syntology's identity for that exact code. It is printed in full (16 hexadecimal characters for every sample in this build), and get_code(code_sha256="…") takes it exactly as printed. An agent calls it on Syntology's MCP service (how to connect); click an identity to copy that call.

Repository labels, per sample. official repository: The archive marks this repository official for the paper. named in the paper: The archive records that the paper mentions this repository; it is not marked official. community (archive-listed): In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper. found in paper text by Syntology: Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted. community: Not in the archive's code links for this paper; a community repository Syntology harvested. Samples are grouped by repository: a repository marked official first, then by how directly the repository is tied to the paper, then in the archive's order; never by how many ran. Licence labels name the repository's licence as recorded at harvest. “Pointer only” means Syntology does not serve that copy's text, for one of six reasons: no licence file was found; the licence was not identified; the licence is recorded as permissive but that copy's record is not marked cleared; the licence is outside the permissive list Syntology serves text under (MIT, Apache-2.0, BSD and similar); this paper's copy was not located, so its licence is not recorded; or this paper's copy is cleared, but the stored copy of the same code that Syntology serves is not. Some licences outside that list permit redistribution, such as WTFPL, and GPL-3.0 under its conditions; they are simply not on the list. Hover a licence label for the reason. File links open the file on GitHub at the default branch, which may have changed since the harvest.

A function or class name links to its page when the name occurs in the harvested code of at least 5 papers; the name is a string, not an identity.

princeton-nlp/benign-data-breaks-safety official repository 14 samples, 7 ran (4 instrument-failed) · 7 unverified
contains_unwanted_words ft_datasets/alpaca_dataset/alpaca_remove_safety.py ran · violated contract no licence file found · pointer only · c4a58b263cc3b3d7 · report fixture: signature text:str/0/str · returned bool · level recorded 2026-09-24
extract_content in 17 papers safety_evaluation/eval_utils/openai_gpt4_judge.py ran fingerprinted no licence file found · pointer only · e28da1bf4c00b938 · report fixture: signature tag:str/0/str, text:str/0/str · returned int · level recorded 2026-09-24
get_credentials safety_evaluation/eval_utils/api_credentials.py ran · our draft was wrong no licence file found · pointer only · f8f41288d948ed1a · report fixture: signature engine:str/0/str, azure:bool/0/bool · returned dict · level recorded 2026-09-24
load_llama_from_config in 8 papers safety_evaluation/eval_utils/model_utils.py ran; the contract check could not run (RAISES) no licence file found · pointer only · b4c052fb373aef07 · report fixture: signature config_path:str/0/str · returned LlamaForCausalLM · level recorded 2026-09-24
load_peft_model in 8 papers safety_evaluation/eval_utils/model_utils.py ran no licence file found · pointer only · 6d5ba8546b6aad2c · report fixture: signature model:callable/0/?, peft_model:str/0/str · returned Mock · level recorded 2026-09-24
question_read safety_evaluation/question_inference.py ran · our draft was wrong no licence file found · pointer only · 4b6ac9bdbe621603 · report fixture: signature text_file:str/0/str · returned list · level recorded 2026-09-24
read_dialogs_from_file in 5 papers safety_evaluation/eval_utils/chat_utils.py ran · our draft was wrong no licence file found · pointer only · 8922fb4b5b7006f7 · report fixture: signature file_path:str/0/str · returned list · level recorded 2026-09-26
contains_unwanted_words ft_datasets/dolly_dataset/dolly_remove_safety.py unverified no licence file found · pointer only · 345f0b1ebadc1536 · report
duo_judge safety_evaluation/eval_utils/openai_gpt4_judge.py unverified no licence file found · pointer only · 30139d8373ac2f31 · report
duo_judge_v2 safety_evaluation/eval_utils/openai_gpt4_judge.py unverified no licence file found · pointer only · c77c19021491eb81 · report
format_tokens in 6 papers safety_evaluation/eval_utils/chat_utils.py unverified no licence file found · pointer only · e084e801c0e634ec · report
load_model in 479 papers safety_evaluation/eval_utils/model_utils.py unverified no licence file found · pointer only · efea256714a4cb99 · report
skip in 14 papers ft_datasets/alpaca_dataset/alpaca_remove_safety.py unverified no licence file found · pointer only · 84cf7fe41466b4b3 · report
skip in 14 papers ft_datasets/dolly_dataset/dolly_remove_safety.py unverified no licence file found · pointer only · a01a8759063203f2 · report

Tasks

Math

Results from the paper archive 2025-07-28

No leaderboard rows for this paper in the archive.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections