activity
20162025
most citedStealing Machine Learning Models via Prediction APIs

733 citations · 899 across the 14 of their papers we have counts for

collaborators
Showing 2024Show all

6 papers · 1 filter

cs.CL2024

Measuring Non-Adversarial Reproduction of Training Data in Large Language Models

Michael Aerni, Javier Rando, Edoardo Debenedetti +3

Large language models memorize parts of their training data. Memorizing short snippets and facts is required to answer questions about the world and to be fluent in any language. B…

cs.LG2024

Evaluating the Robustness of the "Ensemble Everything Everywhere" Defense

Jie Zhang, Christian Schlarmann, Kristina Nikolić +4

Ensemble everything everywhere is a defense to adversarial examples that was recently proposed to make image classifiers robust. This defense works by ensembling a model's intermed…

cs.CV2024

Extracting Training Data from Document-Based VQA Models

Francesco Pinto, Nathalie Rauschmayr, Florian Tramèr +2

Vision-Language Models (VLMs) have made remarkable progress in document-based Visual Question Answering (i.e., responding to queries about the contents of an input document provide…

cs.CR20241 cited

Adversarial Search Engine Optimization for Large Language Models

Fredrik Nestaas, Edoardo Debenedetti, Florian Tramèr

Large Language Models (LLMs) are increasingly used in applications where the model selects from competing third-party content, such as in LLM-powered search engines or chatbot plug…

cs.CR2024

Dataset and Lessons Learned from the 2024 SaTML LLM Capture-the-Flag Competition

Edoardo Debenedetti, Javier Rando, Daniel Paleka +18

Large language model systems face important security risks from maliciously crafted messages that aim to overwrite the system's original instructions or leak private data. To study…

cs.CR2024

Privacy Backdoors: Stealing Data with Corrupted Pretrained Models

Shanglun Feng, Florian Tramèr

Practitioners commonly download pretrained machine learning models from open repositories and finetune them to fit specific applications. We show that this practice introduces a ne…