Matt Fredrikson
Source-listed: Co-founder and Chief Executive Officer; Associate Professor · Gray Swan AI; Carnegie Mellon University
Record snapshot: Sep 3, 2026. Coverage and affiliations may be incomplete or historical.
Selected work
5- Universal and Transferable Adversarial Attacks on Aligned Language ModelsSaved credit: Contributor · 2023Explore people connected to this work →
- Representation Engineering: A Top-Down Approach to AI TransparencySaved credit: Contributor · 2023Explore people connected to this work →
- Improving Alignment and Robustness with Circuit BreakersSaved credit: Contributor · 2024Explore people connected to this work →
- AgentHarm: A Benchmark for Measuring Harmfulness of LLM AgentsSaved credit: Contributor · 2024Explore people connected to this work →
- Model Inversion Attacks that Exploit Confidence Information and Basic CountermeasuresSaved credit: Contributor · 2015Explore people connected to this work →