Kevin Klyman is an AI policy researcher affiliated with the Stanford Center for Research on Foundation Models (Stanford Center for Research on Foundation Models (CRFM)) and the Harvard Belfer Center, whose published work concerns the governance instruments that foundation-model developers apply to their own products.
Research
Klyman is the author of "Acceptable Use Policies for Foundation Models," published on October 22, 2024 in the Proceedings of the Seventh AAAI/ACM Conference on AI, Ethics, and Society. The paper treats acceptable use policies — the legally binding, terms-of-service-adjacent documents that prohibit specific categories of model use — as a class of governance instrument in its own right, and compares them across OpenAI, Anthropic, Google, Meta and Mistral. Its argument is that acceptable use policies are the most widely deployed cross-developer governance artifact and the primary mechanism for category-level usage restrictions, while remaining understudied relative to their reach (Acceptable Use Policies for Foundation Models — Kevin Klyman (AIES 2024)).
He is one of eight listed authors of NIST AI 800-4, alongside Anita Rao, Drew Keller, Neha Kalra, Ryan Steed, Kweku Kwegyir-Aggrey and others; the publication footnotes Kalra, Kwegyir-Aggrey, Klyman and Staheli as former NIST employees (NIST AI 800-4: Challenges to the Monitoring of Deployed AI Systems, Anita Rao).
He served on the review panel for Regulating Under Uncertainty, alongside Nate Persily, Dave Willner, Rishi Bommasani and Dan Ho (Regulating Under Uncertainty: Governance Options for Generative AI (G'sell, 2024)).
Relationships
- related: Stanford Center for Research on Foundation Models (CRFM), Anita Rao, National Institute of Standards and Technology (NIST), Responsible AI Deployment
- supports: Acceptable Use Policies for Foundation Models — Kevin Klyman (AIES 2024), NIST AI 800-4: Challenges to the Monitoring of Deployed AI Systems