AI Governance & Writing
Why I write
Clinicians belong in the room when
these rules are written
A great deal of healthcare AI policy is being shaped right now, in agency guidance, in state legislation, in hospital procurement decisions, and in the quiet contracts that govern data flows between vendors and providers. Much of it is being shaped by people who have never had to make a clinical call under uncertainty, or who have done so but are not also reading the regulation as it is drafted.
Latest from Substack
Recent essays
Substack · Featured
Visit Substack to read the latest essays
I write about artificial intelligence in healthcare from the position of someone who has practiced medicine for two decades, designed a forensic psychiatry application that uses AI, and is now in law school studying how this technology should be governed. The pieces below sit at that intersection.
Selected publications
Where my work has appeared
Daily Journal · California Lawyer · October 18, 2024
Ethical and Privacy Standards in Forensic Psychiatry in the AI Era
Co-author. On the privacy and ethical considerations that arise when artificial intelligence is brought into forensic psychiatric work.
The Gavel · October 2024, Vol. 73, No. 1
Bridging Psychiatry, Law, and AI
On the points where these three fields meet, and on what each owes the others.
San Francisco Chronicle · Health Section · 2013 to 2014
Monthly column on the front page of the Health Section
Twelve issues covering psychiatric topics for a general readership.
International Forensic Psychiatry Conference · 2018
Forensic Psychiatry in the United States of America
Invited presentation at the Serbskiy Institute of Forensic Psychiatry, Moscow.
Hands-on experience
Building a forensic AI application:
Forenpsych
I designed Forenpsych, a secure application built to assist in the preparation and evaluation of competency-to-stand-trial reports for Superior Court proceedings, and I directed the team that built it. Privacy and confidentiality protections were engineered in from the beginning rather than added at the end. The platform received judiciary approval for a pilot program, although the pilot was not ultimately implemented due to administrative constraints. The experience of designing, building, and shepherding a forensic AI tool through legal review has informed how I now think about healthcare AI governance.
What I am paying close attention to
A short, non-exhaustive list of the questions I am working through: how HIPAA and the growing patchwork of state health privacy laws will adapt, or fail to adapt, to AI systems that ingest, infer from, and sometimes generate health information and the consequences of integrating biometric data into these systems.
The second is clinical accountability. When an AI-assisted tool contributes to or executes a diagnostic or treatment decision, where does liability sit, and how should informed consent change in order to reflect that a model was part of the chain of reasoning.
The third is procurement and operational governance. Most of the meaningful decisions about healthcare AI happen inside procurement, vendor contracts, and IT review boards, well before any patient encounter. The rules that shape what is bought or which companies are funded and how they is used are doing more policy work than most public debates acknowledge.
The fourth is mental health technology specifically. I co-founded a mental health technology company and have watched the field from inside as a praciticing psychiatrist for over 25 years. The combination of vulnerable users, lightly regulated tools, and increasingly capable large language models is a category that deserves its own framework.