AppleStorm - Unmasking the Privacy Risks of Apple Intelligence
Privacy analysts uncover prompt leakage and on-device inference data storage vulnerabilities within early releases of Apple Intelligence.
Summary
Overview: Apple Intelligence, Apple’s newest AI product, is designed to enhance productivity with AI while maintaining Apple's focus on user experience and privacy, often highlighting its use of localized models as a key advantage. But how well do these assurances hold up under scrutiny? While Apple emphasizes privacy as a core principle, my findings challenge some of these claims, illustrating the importance of scrutinizing AI-driven assistants before widespread adoption. In this talk, we take a closer l
Key Topics: ook at the data flows within Apple Intelligence, examining how it interacts with user data and the potential security and privacy risks that come with it. Using traffic analysis and OS inspection techniques, we explore what information is accessed, how it moves through the system, and where it gets transmitted. Our findings challenge common security assumptions of Apple, revealing unexpected behaviors and data leaks. From encrypted traffic to data leakage concerns, this presentation will provide practical insights for users and security professionals alike.
Related talks
Creator Stage·41 min
Matt Domko
Details architectural and data pipeline practices used at Rabbit to secure customer identities, anonymize voice data, and control RAG model access.
Briefings·30 min
Yu Wang · Weiteng Chen
As one of the most significant attack surfaces in Operating Systems, the graphics subsystem has always been a focus of the security community.
Briefings·36 min
Andres Riancho · Hillai Ben-Sasson · Ronen Shustin
The overwhelming majority of AI applications run on NVIDIA hardware and software and use NVIDIA tools to containerize and isolate applications running on the same infrastructure.


