Papers › BadChain: Backdoor Chain-of-Thought Prompting for Large Language Models

BadChain: Backdoor Chain-of-Thought Prompting for Large Language Models

20 Jan 2024arXiv:2401.12242archive 2025-07-28

Zhen Xiang, Fengqing Jiang, Zidi Xiong, Bhaskar Ramasubramanian, Radha Poovendran, Bo Li

Large language models (LLMs) are shown to benefit from chain-of-thought (COT) prompting, particularly when tackling tasks that require systematic reasoning processes. On the other hand, COT prompting also poses new vulnerabilities in the form of backdoor attacks, wherein the model will output unintended malicious content under specific backdoor-triggered conditions during inference. Traditional methods for launching backdoor attacks involve either contaminating the training dataset with backdoored instances or directly manipulating the model parameters during deployment. However, these approaches are not practical for commercial LLMs that typically operate via API access. In this paper, we propose BadChain, the first backdoor attack against LLMs employing COT prompting, which does not require access to the training dataset or model parameters and imposes low computational overhead. BadChain leverages the inherent reasoning capabilities of LLMs by inserting a backdoor reasoning step into the sequence of reasoning steps of the model output, thereby altering the final response when a backdoor trigger exists in the query prompt. Empirically, we show the effectiveness of BadChain for two COT strategies across four LLMs (Llama2, GPT-3.5, PaLM2, and GPT-4) and six complex benchmark tasks encompassing arithmetic, commonsense, and symbolic reasoning. Moreover, we show that LLMs endowed with stronger reasoning capabilities exhibit higher susceptibility to BadChain, exemplified by a high average attack success rate of 97.0% across the six benchmark tasks on GPT-4. Finally, we propose two defenses based on shuffling and demonstrate their overall ineffectiveness against BadChain. Therefore, BadChain remains a severe threat to LLMs, underscoring the urgency for the development of robust and effective future defenses.

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 15 code samples harvested from 1 repository linked to this paper: 13 ran (of which 0 constructed an object rather than computing a result; 10 with no instrument failure: 0 honoured, 0 violated, 10 with no contract checked; 3 where Syntology's instrument failed) · 2 unverified. Where Syntology's instrument failed: 3 where our contract draft was wrong; failures of Syntology's instrument, not of the code.

By repository: official repository: 15 samples from 1 repository, 13 ran (10 with no instrument failure; 3 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 on 2026-09-24 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

django-jiang/badchain officialmentioned in papermentioned on GitHubMITSyntology: 15 samples, 13 ran (3 instrument-failed) · 2 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

15 samples harvested: 13 ran (of which 0 constructed an object rather than computing a result; 10 with no instrument failure: 0 honoured, 0 violated, 10 with no contract checked; 3 where Syntology's instrument failed) · 2 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.

3ran · our draft was wrong
10ran
2unverified

Licence: 0 of the 15 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 django-jiang/badchain. “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.

django-jiang/badchain official repository 15 samples, 13 ran (3 instrument-failed) · 2 unverified
bd_embed attack.py ran fingerprinted MIT (permissive) · ba380522faabf52a · report fixture: signature q:str/0/str, trigger_type:str/0/str, trigger:str/0/str, position:str/0/str · returned str · level recorded 2026-09-24
check_step ans_eval/MATH.py ran MIT (permissive) · beecceaab5326635 · report fixture: signature doc:str/0/str, pool:list/1/str · returned tuple · level recorded 2026-09-24
defense defense.py ran MIT (permissive) · a4fa464a1b6e7443 · report fixture: signature prompt:str/0/str, strategy:str/0/str · returned str · level recorded 2026-09-24
eval_answer_perturb_check_step ans_eval/MATH.py ran MIT (permissive) · d5a7cb2951759b5b · report fixture: signature pred_str:str/0/str, ans_str:str/0/str, factor:int/0/int · returned bool · level recorded 2026-09-24
find_answer ans_eval/MATH.py ran · our draft was wrong MIT (permissive) · 4b2d5b42a427bb0b · report fixture: signature s:str/0/str · returned str · level recorded 2026-09-24
find_step ans_eval/ASDiv.py ran MIT (permissive) · bf24fa1953f8d2fe · report fixture: signature doc:str/0/str, pool:list/1/str · returned tuple · level recorded 2026-09-24
shift_answer ans_eval/csqa.py ran MIT (permissive) · 595f4302a96ec9db · report fixture: signature ans:str/0/str, shift:int/0/int, mode:NoneType/0/? · returned str · level recorded 2026-09-24
shuffle_lines defense.py ran · our draft was wrong MIT (permissive) · 5f8b2379ca5acce5 · report fixture: signature doc:str/0/str · returned str · level recorded 2026-09-24
shuffle_words defense.py ran · our draft was wrong MIT (permissive) · 1f22e4fee93daab7 · report fixture: signature doc:str/0/str · returned str · level recorded 2026-09-24
step_exist ans_eval/ASDiv.py ran MIT (permissive) · ff778bcd029a43bf · report fixture: signature pred_str:str/0/str, ans_str:str/0/str, factor:int/0/int · returned bool · level recorded 2026-09-24
step_exists ans_eval/csqa.py ran MIT (permissive) · c579e0bc1e4f691d · report fixture: signature pred_str:str/0/str, shift:int/0/int · returned bool · level recorded 2026-09-24
test_answer in 6 papers ans_eval/ASDiv.py ran MIT (permissive) · 91bd5c1829919390 · report fixture: signature pred_str:str/0/str, ans_str:str/0/str, factor:float/0/float · returned bool · level recorded 2026-09-24
test_answer in 6 papers ans_eval/csqa.py ran MIT (permissive) · b91f9f3a9644f423 · report fixture: signature pred_str:str/0/str, ans_str:str/0/str, shift:int/0/int · returned bool · level recorded 2026-09-24
parse_pred_ans in 5 papers ans_eval/letter.py unverified MIT (permissive) · 4d8f67ee98a17674 · report
parse_pred_ans in 5 papers ans_eval/strategyqa.py unverified MIT (permissive) · f03e3e5ae911035c · report

Tasks

Backdoor Attack

Results from the paper archive 2025-07-28

No leaderboard rows for this paper in the archive.

Methods

Absolute Position EncodingsAdamAttentionAttention DropoutBPECoT PromptingCosine AnnealingDense ConnectionsDropoutGPT-3GPT-4Label SmoothingLayer NormalizationLinear LayerLinear Warmup With Cosine AnnealingMulti-Head AttentionPosition-Wise Feed-Forward LayerResidual ConnectionSoftmaxTransformerWeight Decay

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