ML&|Sec Feed

email2phonenumber — инструмент OSINT, позволяющий получить номер телефона цели, просто зная ее адрес электронной почты 4️⃣ GitHub

Introducing MLSecOps by Adversa in Conf42 https://conf42.github.io/static/slides/Conf42%20DevSecOps%202022%20-%20Eugene%20Neelou.pdf

GUARDIAN: A Multi-Tiered Defense Architecture for Thwarting Prompt Injection Attacks on LLMs https://www.scirp.org/journal/paperinformation?paperid=130663

Practical Attacks on Machine Learning: A Case Study on Adversarial Windows Malware https://arxiv.org/pdf/2207.05548

Stealing Part of a Production Language Model Carlini et al. https://not-just-memorization.github.io/partial-model-stealing.html

Model Inversion Attacks that Exploit Confidence Information and Basic Countermeasures https://rist.tech.cornell.edu/papers/mi-ccs.pdf

Extracting Training Data from Large Language Models https://arxiv.org/pdf/2012.07805

Not what you’ve signed up for: Compromising Real-World LLM-Integrated Applications with Indirect Prompt Injection https://arxiv.org/pdf/2302.12173

Jailbreak Attacks and Defenses Against Large Language Models: A Survey https://arxiv.org/pdf/2407.04295

AI Security Glossary | Словарь терминов по безопасности ИИ https://www.nightfall.ai/ai-security-101

Efficient Defenses Against Adversarial Attacks - Bounded RELU - Gaussian Data Augmentation https://arxiv.org/pdf/1707.06728

Старые, но неплохие статьи Безопасность алгоритмов машинного обучения. Атаки с использованием Python Безопасность алгоритмов машинного обучения. Защита и тестирование моделей с использованием Python

On Evaluating Adversarial Robustness https://arxiv.org/pdf/1902.06705

Thermometer Encoding: One-Hot Way To Resist Adversarial Examples https://openreview.net/pdf?id=S18Su--CW