M. Bilgehan Ertan

M. Bilgehan Ertan is an individual researcher specializing in AI safety, privacy, and machine learning. He is pursuing a PhD at CWI (Centrum Wiskunde & Informatica) in Amsterdam, under the supervision of Professor Marten van Dijk. His background includes a bachelor’s degree in Computer Science and Engineering from Sabancı University and a cum laude master’s degree in Computer Security from Vrije Universiteit Amsterdam. He has also worked as a Security Researcher and Developer at PRODAFT.

Ertan’s research aims to make AI systems safer, more secure, and privacy-preserving while maintaining practical utility. His work examines information leakage and adversarial behavior in modern AI, using frameworks such as differential privacy, f-DP, and PAC privacy. He studies membership inference and adversarial attacks on large language models, and collaborates with Professor Srini Devadas at MIT on certified defenses against data poisoning in LLM post-training and the safety of agentic systems. His publications explore privacy-aware model fine-tuning, DP-SGD trade-offs, privacy guarantees against inference attacks, and the limits of membership inference for copyright auditing.