About
Matthew works at the seam of law and machine learning: making AI systems legible and accountable enough to trust in high-stakes settings like M&A, forensic investigations, and the courtroom. He's currently researching how language models internally represent legal concepts (BlueDot grant), after technical AI safety work at TARA and AI evaluation work in legal and forensic settings. He's interested in interpretability as a governance tool, and in getting the technical and policy worlds talking. Law + CS (Hons), UQ.