Copertina

Poisoning Web-Scale Training Datasets is Practical

Fuente: arXiv

Salvato in:

Dettagli Bibliografici
Autori principali:	Carlini, Nicholas, Jagielski, Matthew, Choquette-Choo, Christopher A., Paleka, Daniel, Pearce, Will, Anderson, Hyrum, Terzis, Andreas, Thomas, Kurt, Tramèr, Florian
Natura:	Preprint
Pubblicazione:	2023
Soggetti:	Cryptography and Security Machine Learning
Accesso online:	Acceder al recurso
Tags:	Aggiungi Tag Nessun Tag, puoi essere il primo ad aggiungerne!!

Documenti analoghi

Privacy Side Channels in Machine Learning Systems
di: Debenedetti, Edoardo, et al.
Pubblicazione: (2023)

LLMs unlock new paths to monetizing exploits
di: Carlini, Nicholas, et al.
Pubblicazione: (2025)

Auditing Private Prediction
di: Chadha, Karan, et al.
Pubblicazione: (2024)

Position: Considerations for Differentially Private Learning with Large-Scale Public Pretraining
di: Tramèr, Florian, et al.
Pubblicazione: (2022)

Large-scale online deanonymization with LLMs
di: Lermen, Simon, et al.
Pubblicazione: (2026)

Persistent Pre-Training Poisoning of LLMs
di: Zhang, Yiming, et al.
Pubblicazione: (2024)

The Last Iterate Advantage: Empirical Auditing and Principled Heuristic Analysis of Differentially Private SGD
di: Steinke, Thomas, et al.
Pubblicazione: (2024)

Evading Black-box Classifiers Without Breaking Eggs
di: Debenedetti, Edoardo, et al.
Pubblicazione: (2023)

Privacy Ripple Effects from Adding or Removing Personal Information in Language Model Training
di: Borkar, Jaydeep, et al.
Pubblicazione: (2025)

Adversarial Perturbations Cannot Reliably Protect Artists From Generative AI
di: Hönig, Robert, et al.
Pubblicazione: (2024)

Stealing Part of a Production Language Model
di: Carlini, Nicholas, et al.
Pubblicazione: (2024)

Are aligned neural networks adversarially aligned?
di: Carlini, Nicholas, et al.
Pubblicazione: (2023)

Exploring and Mitigating Adversarial Manipulation of Voting-Based Leaderboards
di: Huang, Yangsibo, et al.
Pubblicazione: (2025)

Adversarial ML Problems Are Getting Harder to Solve and to Evaluate
di: Rando, Javier, et al.
Pubblicazione: (2025)

Defeating Prompt Injections by Design
di: Debenedetti, Edoardo, et al.
Pubblicazione: (2025)

Language Models May Verbatim Complete Text They Were Not Explicitly Trained On
di: Liu, Ken Ziyu, et al.
Pubblicazione: (2025)

Universal Jailbreak Backdoors from Poisoned Human Feedback
di: Rando, Javier, et al.
Pubblicazione: (2023)

AutoAdvExBench: Benchmarking autonomous exploitation of adversarial example defenses
di: Carlini, Nicholas, et al.
Pubblicazione: (2025)

Near Exact Privacy Amplification for Matrix Mechanisms
di: Choquette-Choo, Christopher A., et al.
Pubblicazione: (2024)

Optimal Rates for $O(1)$-Smooth DP-SCO with a Single Epoch and Large Batches
di: Choquette-Choo, Christopher A., et al.
Pubblicazione: (2024)

Privacy Amplification for Matrix Mechanisms
di: Choquette-Choo, Christopher A., et al.
Pubblicazione: (2023)

Evaluating the Robustness of a Production Malware Detection System to Transferable Adversarial Attacks
di: Nasr, Milad, et al.
Pubblicazione: (2025)

Evaluating the Robustness of the "Ensemble Everything Everywhere" Defense
di: Zhang, Jie, et al.
Pubblicazione: (2024)

Query-Based Adversarial Prompt Generation
di: Hayase, Jonathan, et al.
Pubblicazione: (2024)

The Attacker Moves Second: Stronger Adaptive Attacks Bypass Defenses Against Llm Jailbreaks and Prompt Injections
di: Nasr, Milad, et al.
Pubblicazione: (2025)

Phantom: General Backdoor Attacks on Retrieval Augmented Language Generation
di: Chaudhari, Harsh, et al.
Pubblicazione: (2024)

Privacy Backdoors: Stealing Data with Corrupted Pretrained Models
di: Feng, Shanglun, et al.
Pubblicazione: (2024)

Covert Attacks on Machine Learning Training in Passively Secure MPC
di: Jagielski, Matthew, et al.
Pubblicazione: (2025)

Membership Inference Attacks Cannot Prove that a Model Was Trained On Your Data
di: Zhang, Jie, et al.
Pubblicazione: (2024)

Thought-Transfer: Indirect Targeted Poisoning Attacks on Chain-of-Thought Reasoning Models
di: Chaudhari, Harsh, et al.
Pubblicazione: (2026)

Certified Robustness to Clean-Label Poisoning Using Diffusion Denoising
di: Hong, Sanghyun, et al.
Pubblicazione: (2024)

Cutting through buggy adversarial example defenses: fixing 1 line of code breaks Sabre
di: Carlini, Nicholas
Pubblicazione: (2024)

Privacy Backdoors: Enhancing Membership Inference through Poisoning Pre-trained Models
di: Wen, Yuxin, et al.
Pubblicazione: (2024)

Adversarial Search Engine Optimization for Large Language Models
di: Nestaas, Fredrik, et al.
Pubblicazione: (2024)

Evaluations of Machine Learning Privacy Defenses are Misleading
di: Aerni, Michael, et al.
Pubblicazione: (2024)

Lessons from Defending Gemini Against Indirect Prompt Injections
di: Shi, Chongyang, et al.
Pubblicazione: (2025)

Mellivora Capensis: A Backdoor-Free Training Framework on the Poisoned Dataset without Auxiliary Data
di: Pu, Yuwen, et al.
Pubblicazione: (2024)

On Evaluating the Durability of Safeguards for Open-Weight LLMs
di: Qi, Xiangyu, et al.
Pubblicazione: (2024)

EMBER2024 -- A Benchmark Dataset for Holistic Evaluation of Malware Classifiers
di: Joyce, Robert J., et al.
Pubblicazione: (2025)

Privacy Auditing of Large Language Models
di: Panda, Ashwinee, et al.
Pubblicazione: (2025)