PhD Position | Information-theoretic foundations of memory poisoning

We are looking for a highly talented and motivated Ph.D. candidate to work on information-theoretic foundations of memory poisoning at Linköping University

The project, funded by the Swedish Research Council (VR), aims to develop a rigorous understanding of how malicious information injected into an AI agent’s memory can influence its reasoning and decision-making.

It is suitable for a candidate who is interested in developing rigorous foundations while also validating ideas through selected proof-of-concept attacks, defenses, and experiments with LLM-based agents.

Prior experience in AI security, privacy, or LLMs is beneficial but not required. We especially welcome strong applicants from backgrounds such as information theory, communication theory, signal processing, optimization and learning, applied probability and statistics, cybersecurity, and distributed systems.


Visit: https://liu.se/en/work-at-liu/vacancies/29341