Topics
Browse posts by category and tag — every topic we cover, with the latest pieces under each.
Tags
- #adversarial-ml 20
- #ml-security 13
- #adversarial-examples 6
- #evasion-attacks 4
- #privacy 4
- #adversarial-robustness 3
- #backdoor-attacks 3
- #data-poisoning 3
- #llm-security 3
- #tooling 3
- #vision-language-models 3
- #adversarial-training 2
- #ai-red-teaming 2
- #clip-attacks 2
- #image-classifiers 2
- #membership-inference 2
- #model-extraction 2
- #model-inversion 2
- #model-testing 2
- #multimodal-adversarial-attacks 2
- #red-team 2
- #robustness 2
- #robustness-evaluation 2
- #transferability 2
- #universal-adversarial-perturbations 2
- #adaptive-attacks 1
- #adversarial-defense 1
- #adversarial-machine-learning 1
- #adversarial-nlp 1
- #adversarial-patches 1
- #alignment 1
- #api-security 1
- #autoattack 1
- #autonomous-vehicles 1
- #black-box-attacks 1
- #carlini-wagner 1
- #certified-robustness 1
- #differential-privacy 1
- #embedding-inversion 1
- #evaluation 1
- #evasion 1
- #feature-space-attacks 1
- #federated-learning 1
- #fgsm 1
- #formal-verification 1
- #foundation-models 1
- #garak 1
- #gcg 1
- #gdpr 1
- #gradient-inversion 1
- #gradient-masking 1
- #graph-neural-networks 1
- #input-agnostic-attacks 1
- #jailbreaking 1
- #mart 1
- #memorization 1
- #metrics 1
- #ml-privacy 1
- #mlaas 1
- #mlsecops 1
- #model-hardening 1
- #model-reprogramming 1
- #model-stealing 1
- #network-security 1
- #neural-networks 1
- #nids 1
- #nlp 1
- #object-detection 1
- #optimization-attacks 1
- #pgd 1
- #physical-adversarial-attacks 1
- #production-ml 1
- #pyrit 1
- #randomized-smoothing 1
- #robustbench 1
- #robustness-accuracy-tradeoff 1
- #text-attacks 1
- #threat-modeling 1
- #trades 1
- #training-data 1
- #training-data-extraction 1
- #training-data-privacy 1
- #transformers 1
- #trojan-ml 1
- #unforeseen-attacks 1
- #vec2text 1
- #vector-store 1
- #visual-adversarial-examples 1
Categories
attacks 11 posts
- CLIP Adversarial Attacks: One Perturbation, Every ModelCLIP is the shared vision encoder under most multimodal systems, which is why one universal perturbation now transfers across encoders, tasks and domains.
- PLAA: A 92.78% NIDS Evasion Rate and Feature-Space AttacksPLAA builds adversarial network traffic at the packet level instead of the flow level, reaching a 92.78% evasion rate against deep-learning NIDS.
- Adversarial Attacks on Vision-Language Models: CLIP, LLaVA, GPT-4Vision-language models widen the adversarial attack surface: crafted images can steer text output, carry typographic payloads, and jailbreak the model.
- Adversarial Patch Attacks: Physical Perturbations That Fool MLAdversarial patches are large, visible, localized perturbations designed to survive physical-world conditions — printing, lighting, and camera optics.
- Universal Adversarial Perturbations: One Vector That Fools InputsUniversal adversarial perturbations are input-agnostic: one crafted noise vector causes misclassification across most inputs and transfers between models.
- Adversarial Robustness in NLP: Why Text Attacks Are DifferentDiscrete input spaces, semantic constraints, and human-perceptibility rules change what counts as an adversarial example in text.
defenses 5 posts
- GraphRP: A Structure-Aware Defense Against GNN Model ExtractionGraphRP repurposes model reprogramming as a proactive defense against GNN model extraction, avoiding the Euclidean bias that image-borne defenses import.
- Data Poisoning Attack Detection Methods That Actually WorkA practitioner's rundown of data poisoning attack detection methods: spectral signatures, activation clustering, and what OWASP and NIST recommend.
- Adversarial Training Best Practices That Survive EvaluationA practitioner's guide to adversarial training: pick a threat model, run PGD-AT correctly, catch robust overfitting, and evaluate with AutoAttack.
- Adversarial Training Methods: PGD-AT, TRADES, and MARTAdversarial training is the most defensible empirical robustness method, but 'adversarial training' isn't one thing.
- Certified Robustness via Randomized Smoothing: What It GuaranteesRandomized smoothing gives a provable robustness radius. What that certificate actually covers, where it breaks at realistic epsilon, and when to use it.
privacy 4 posts
- Embedding Inversion: Reconstructing Text From VectorsEmbedding inversion recovers the original text from a model's embedding vectors, breaking the assumption that stored embeddings are privacy-safe.
- Model Inversion Attacks: Reconstructing Training Data from OutputFrom Fredrikson's pharmacogenetics exploit to gradient inversion, model inversion attacks recover private training data most ML teams assume is safe.
- Training Data Extraction from LLMs: The Carlini Results ExplainedCarlini et al. demonstrated verbatim extraction of training data from GPT-2. The results have been widely misread. What the extraction numbers actually show.
- Membership Inference Attacks: What Works on Production ML APIsShokri et al.'s shadow-model attack is the canonical reference, but the gap between the paper's threat model and a real rate-limited API is wide.
evaluation 2 posts
- Testing Robustness Against Unforeseen Adversaries: The UAR MetricOpenAI's Unforeseen Attack Robustness metric scores classifiers against adversarial perturbations they never trained on. What it measures and misses.
- Evaluating Adversarial Robustness Without Fooling YourselfMost defenses that claim robustness are later broken — not because the idea was bad, but because the evaluation was. Adaptive attacks are non-negotiable.
primer 2 posts
- Adversarial Examples vs. Data Poisoning: Timing Is EverythingAdversarial examples attack a deployed model at inference; data poisoning attacks the model before it is deployed. Different timing, different defenses.
- Membership Inference vs. Model Inversion: Privacy AttacksMembership inference asks 'was this sample in the training set?' Model inversion asks 'what samples were in the training set?
tooling 2 posts
- Adversarial Attack Libraries: ART, Foolbox, torchattacksART, Foolbox, torchattacks, CleverHans and TextAttack compared on scope, framework support, attack coverage, licence and repository maintenance.
- Best AI Red Teaming Tools for ML Models in 2026A practitioner's guide to the tools used to red team machine learning models: PyRIT, garak, ART, ModelScan, and Giskard, and where each fits a pipeline.