CodeCloak: A Method for Evaluating and Mitigating Code Leakage by LLM Code Assistants

no code implementations • 13 Apr 2024 • Amit Finkman, Eden Bar-Kochva, Avishag Shapira, Dudu Mimran, Yuval Elovici, Asaf Shabtai

While beneficial, these tools might inadvertently expose the developer's proprietary code to the code assistant service provider during the development process.

Adversarial Machine Learning Threat Analysis and Remediation in Open Radio Access Network (O-RAN)

no code implementations • 16 Jan 2022 • Edan Habler, Ron Bitton, Dan Avraham, Dudu Mimran, Eitan Klevansky, Oleg Brodt, Heiko Lehmann, Yuval Elovici, Asaf Shabtai

Next, we explore the various AML threats associated with O-RAN and review a large number of attacks that can be performed to realize these threats and demonstrate an AML attack on a traffic steering model.

Anomaly Detection BIG-bench Machine Learning