Papers › AdvCLIP: Downstream-agnostic Adversarial Examples in Multimodal Contrastive Learning

AdvCLIP: Downstream-agnostic Adversarial Examples in Multimodal Contrastive Learning

14 Aug 2023arXiv:2308.07026archive 2025-07-28

Ziqi Zhou, Shengshan Hu, Minghui Li, Hangtao Zhang, Yechao Zhang, Hai Jin

Multimodal contrastive learning aims to train a general-purpose feature extractor, such as CLIP, on vast amounts of raw, unlabeled paired image-text data. This can greatly benefit various complex downstream tasks, including cross-modal image-text retrieval and image classification. Despite its promising prospect, the security issue of cross-modal pre-trained encoder has not been fully explored yet, especially when the pre-trained encoder is publicly available for commercial use. In this work, we propose AdvCLIP, the first attack framework for generating downstream-agnostic adversarial examples based on cross-modal pre-trained encoders. AdvCLIP aims to construct a universal adversarial patch for a set of natural images that can fool all the downstream tasks inheriting the victim cross-modal pre-trained encoder. To address the challenges of heterogeneity between different modalities and unknown downstream tasks, we first build a topological graph structure to capture the relevant positions between target samples and their neighbors. Then, we design a topology-deviation based generative adversarial network to generate a universal adversarial patch. By adding the patch to images, we minimize their embeddings similarity to different modality and perturb the sample distribution in the feature space, achieving unviersal non-targeted attacks. Our results demonstrate the excellent attack performance of AdvCLIP on two types of downstream tasks across eight datasets. We also tailor three popular defenses to mitigate AdvCLIP, highlighting the need for new defense mechanisms to defend cross-modal pre-trained encoders.

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 16 code samples harvested from 1 repository linked to this paper: 15 ran (of which 0 constructed an object rather than computing a result; 11 with no instrument failure: 1 honoured, 1 violated, 9 with no contract checked; 4 where Syntology's instrument failed) · 1 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: 16 samples from 1 repository, 15 ran (11 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 2026-09-22 and 2026-09-25 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

cgcl-codes/advclip officialmentioned in paperpytorchMITSyntology: 16 samples, 15 ran (4 instrument-failed) · 1 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

16 samples harvested: 15 ran (of which 0 constructed an object rather than computing a result; 11 with no instrument failure: 1 honoured, 1 violated, 9 with no contract checked; 4 where Syntology's instrument failed) · 1 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 · honoured contract
1ran · violated contract
3ran · our draft was wrong
1ran; the contract check could not run
9ran
1unverified

Licence: 5 of the 16 samples are pointer only, meaning Syntology does not serve that copy's text. For 5 of them, this paper's copy is under a permissive licence and cleared, but the stored copy of the same code that Syntology serves is not; the licence label's tooltip gives Syntology's recorded reason. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from cgcl-codes/advclip. “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. A census tag appears when Syntology's census ran implementations with that name on one shared input per bucket: it gives how many were compared, from how many papers, and how many distinct outputs they produced, and names this sample's bucket when this sample was one of them.

cgcl-codes/advclip official repository 16 samples, 15 ran (4 instrument-failed) · 1 unverified
CE in 7 papers utils/metrics.py ran fingerprinted MIT (permissive) here; the stored copy is pointer only · 0daa5f9fe5b85ae3 · report fixture: signature P:Tensor/2/float, Q:Tensor/2/float, mask:Tensor/2/float · returned Tensor · level recorded 2026-09-24
KL in 20 papers utils/metrics.py ran fingerprinted MIT (permissive) here; the stored copy is pointer only · fb0541d3902f7b0c · report fixture: signature P:Tensor/2/float, Q:Tensor/2/float, mask:Tensor/2/float · returned Tensor · level recorded 2026-09-24
accuracy in 872 papers
census: 8 implementations from 7 papers, 7 distinct outputs across 6 buckets, one shared input per bucket
train_downstream_solo.py ran; the contract check could not run (RAISES) · metamorphic tier: invariant MIT (permissive) · 1e369e39c90cada6 · report fixture: signature output:Tensor/2/float, target:Tensor/1/int, topk:tuple/1/int · returned list · level recorded 2026-09-22
clamp_patch utils/patch_utils.py ran MIT (permissive) · 3a01df6a1dc9521d · report fixture: signature args:dict/0/?, patch:Tensor/4/float · returned Tensor · level recorded 2026-09-24
fx_calc_map_label utils/evaluate.py ran MIT (permissive) · 299ed83aade55bae · report fixture: signature image:ndarray/2/float, text:ndarray/2/float, label:ndarray/1/int, k:int/0/int, dist_method:str/0/str · returned float64 · level recorded 2026-09-24
fx_calc_recall utils/evaluate.py ran MIT (permissive) · 19a8a77e50cab4fd · report fixture: signature image:ndarray/2/float, text:ndarray/2/float, label:ndarray/1/int, k:int/0/int, dist_method:str/0/str · returned tuple · level recorded 2026-09-24
gelu in 308 papers
census: 43 implementations from 74 papers, 6 distinct outputs on one shared input this sample: bucket 1 of 1, arg 1: rank 2, kind float, dtype float32
utils/model.py ran · honoured contract fingerprinted MIT (permissive) · c2f0705ed8e6f7e6 · report fixture: signature x:Tensor/2/float · returned Tensor · level recorded 2026-09-22
info_nce in 21 papers
census: 6 implementations from 5 papers, 3 distinct outputs across 2 buckets, one shared input per bucket
utils/nce.py ran · our draft was wrong MIT (permissive) here; the stored copy is pointer only · 54b175c46812e410 · report fixture: signature query:Tensor/2/float, positive_key:Tensor/2/float, negative_keys:Tensor/2/float, temperature:float/0/float, reduction:str/0/str, negative_mode:str/0/str · returned Tensor · level recorded 2026-09-25
l1norm in 14 papers
census: 7 implementations from 7 papers, 2 distinct outputs on one shared input this sample: bucket 1 of 1, arg 1: rank 2, kind float, dtype float32
utils/model.py ran · our draft was wrong fingerprinted MIT (permissive) here; the stored copy is pointer only · 9bf455ab09372ab4 · report fixture: signature X:Tensor/2/float, dim:int/0/int, eps:float/0/float · returned Tensor · level recorded 2026-09-22
l2norm in 84 papers
census: 12 implementations from 20 papers, 3 distinct outputs on one shared input this sample: bucket 1 of 1, arg 1: rank 2, kind float, dtype float32
utils/model.py ran · our draft was wrong fingerprinted MIT (permissive) · be7c4a676b9f1e19 · report fixture: signature X:Tensor/2/float, dim:int/0/int, eps:float/0/float · returned Tensor · level recorded 2026-09-22
make_print_to_file utils/evaluate.py ran MIT (permissive) · e1d898daa31e30ff · report fixture: signature path:str/0/str · returned str · level recorded 2026-09-24
mask_generation utils/patch_utils.py ran MIT (permissive) · d5a0abdbf17edceb · report fixture: signature args:Args/0/?, patch:ndarray/3/float · returned tuple · level recorded 2026-09-24
patch_initialization utils/patch_utils.py ran MIT (permissive) · 81e84201408ca9e3 · report fixture: signature args:Args/0/?, patch_type:str/0/str · returned ndarray · level recorded 2026-09-24
test in 377 papers train_downstream_solo.py ran MIT (permissive) · f6dd2b488b69305c · report fixture: signature encoder:callable/0/?, classifier:callable/0/?, test_loader:list/1/tuple, device:device/0/? · returned tuple · level recorded 2026-09-24
transpose in 53 papers
census: 6 implementations from 13 papers, 3 distinct outputs across 2 buckets, one shared input per bucket this sample: bucket 1 of 2, arg 1: rank 4, kind float, dtype float32
utils/nce.py ran · violated contract fingerprinted MIT (permissive) · 55f1419d4d8c483c · report fixture: signature x:Tensor/4/float · returned Tensor · level recorded 2026-09-22
algorithm2 utils/metrics.py unverified MIT (permissive) here; the stored copy is pointer only · 2c4ef32312957719 · report

Tasks

Contrastive LearningImage ClassificationImage-text RetrievalText Retrievalimage-classification

1 archive task tag without a task page not shown.

Results from the paper archive 2025-07-28

No leaderboard rows for this paper in the archive.

Methods

CLIPContrastive Learning

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