Technical AI Governance Research

The Technical AI Governance Research (TAIGR) Lab studies how AI systems behave and affect the world, investigates successes and failures, and develops tools to inform AI governance.

The Technical AI Governance Research (TAIGR) Lab, led by Co-PIs Anka Reuel and Stephen Casper, works to close gaps in understanding and tooling for governing AI. We develop rigorous ways to evaluate the capabilities of AI, study its diffuse impacts, and examine which governance interventions work and why. By connecting evaluation with real-world use and incidents, we aim to create a feedback loop that helps AI governance adapt as the technology evolves.

MIT Technology Review: We still don’t know how people are really using AI
Coverage of the AI Observatory’s finding that company usage reports skew toward work cases: applying Anthropic’s method to the Observatory’s data would have filtered out 48% of conversations.

The Washington Post: New data shows people are getting personal with AI
The Washington Post makes the case for independent, cross-provider measurement of AI use, and what policymakers currently cannot see based on the AI Observatory

The AI Observatory
A public measure of real-world AI use: 85,633 conversational turns across seven consented datasets, aggregated into an independent picture of how people actually use AI systems.

Stephen Casper
Co-principal investigator

I’m Stephen Casper, but most people call me Cas. I am a computer scientist and an Assistant Professor at the Harvard Kennedy School.

Prior to joining Harvard, I completed my PhD at MIT and did a research residency with the UK AI Security Institute. I am a writer for the International AI Safety Report and a lead writer for the Singapore Consensus. I also work with MATS, CBAI, ERA, and GovAI to lead research streams. I never have and never will work for big tech and have never signed a non-disparagement agreement. 

Anka Reuel
Co-principal investigator

I'm Anka and I'm currently wrapping up my Computer Science PhD at Stanford University before joining the Harvard Kennedy School as an Assistant Professor (with a secondary appointment in Computer Science) in January 2027. 

I served as a vice chair for the European Commission in the EU's General-Purpose AI Code of Practice Process and was a 2024-25 Technology & Geopolitics Fellow at the Belfer Center at Harvard Kennedy School. I was the Responsible AI Lead for the 2024 and 2025 Stanford AI Index and the Lead AI Researcher for the Stanford Emerging Technology Review 2024 and 2025.

The inaugural TAIGR Conference will happen in 2027.

Further information will follow.