University of Cambridge > Talks.cam > Cambridge ML Systems Seminar Series > Navigating Privacy Risks in Language Models

  Browse A-Z

Log in

University Account External (via Google)

🔐 Log In

Information on

Subscribing to talksdetails Finding a talkdetails Adding a talkdetails Disseminating talksdetails Help and Documentationdetails

ℹī¸ Information

Navigating Privacy Risks in Language Models

Download to your calendar using vCal

If you have a question about this talk, please contact Nic Lane .

The emergence of large language models (LLMs) presents significant opportunities in content generation, question answering, and information retrieval. Nonetheless, training, fine-tuning, and deploying these models entails privacy risks. This talk will address these risks, outlining privacy principles inspired by known LLM vulnerabilities when handling user data. We demonstrate how techniques like federated learning and user-level differential privacy (DP) can systematically mitigate many of these risks at the cost of increased computation. In scenarios where only moderate-to-weak user-level DP is achievable, we propose a strong (task-and-model-agnostic) membership inference attack that allows us to quantify risk by estimating the actual privacy leakage (empirical epsilon) accurately in a single training run. The talk will conclude with a few projections and compelling research directions.

This talk is part of the Cambridge ML Systems Seminar Series series.

This talk is included in these lists:

Note that ex-directory lists are not shown.

Other lists

Hybrid: in-person in Cambridge & online via Teams Average on Sundays: 7478.48148148 kWh Average on weekdays: 9017.3707483 kWh

Other talks

Title and abstract tba Clonal haematopoiesis, early detection and pre-leukemia to leukemia development Professor Philip Rosenstiel, W3 Professor of Clinical Molecular Biology, Director, Institute of Clinical Molecular Biology (IKMB), Kiel University Am I a Member? Auditing Private Machine Learning Shape modes of CP^1 vortices Title TBC
 

Š 2006-2025 Talks.cam, University of Cambridge. Contact Us | Help and Documentation | Privacy and Publicity