AI Safety Group @ UChicago XLab logoAI Safety Group@ UChicago XLab

Fellowship Syllabus

Policy and Governance

This track builds a working map of AI policy: the key players and institutions, the functions of government that shape AI, and the policy levers available. Fellows survey the frameworks in place today, explore pressing topics and issues, as well as proposed responses to address them. Fellows conclude the program by drafting and defending their own policy briefs. No prior policy experience is required — we welcome students from technical backgrounds looking to build policy literacy, as well as those in the social sciences and humanities engaging with AI-specific governance challenges.

Session 01

Introduction to AI Policy and Governance

What might transformative AI look like, and what would it take for its governance to go well? This session builds a shared picture of where AI is headed — from scaling trends to the possibility of an intelligence explosion — and surveys high-level plans for steering it, from theories of victory to staged responses to misalignment risk.

Session 02

Understanding the Landscape

How is AI policy researched and implemented, and what proposals exist? This session introduces the key policy levers available to government, with fellows splitting into groups to cover one lever each, then turns to concrete proposals from industry, Congress, and policy writers.

Key policy levers

  • Federal R&D Funding

    How federal research funding shapes the direction of emerging technologies, and how it gets allocated.

  • Regulatory Policy

    How agencies write and enforce rules, and where regulation fits in technology policy.

  • Technical Standards and Evaluations

    The role of standards bodies and model evaluations in governing AI.

  • Export Controls

    How export controls on chips and other technologies work, and how they are used in AI policy.

Proposals

  • Policy on the AI Exponential

    Dario Amodei argues that exponential AI progress has outpaced the policy process, and proposes mandatory frontier model testing, labor-market preparation, and a coalition of democracies.

  • Sen. Bernie Sanders at the National Press Club: Q&A13:09–25:30

    Senator Bernie Sanders takes questions at a National Press Club headliner event.

  • Be It Enacted

    Dean W. Ball drafts a federal AI bill pairing transparency requirements for frontier developers with a temporary preemption of state AI laws.

Session 03

Current Policy and Governance

What rules already govern frontier AI? This session surveys US executive policy, the frontier AI safety bills in California, New York, and Illinois, the safety frameworks published by leading AI developers, and the EU AI Act. For the state bills and developer frameworks, fellows split into groups to read one each, then come together to compare where they converge and where they differ.

Executive policy

State legislation

  • California SB 53

    California's Transparency in Frontier Artificial Intelligence Act, requiring large frontier developers to publish safety frameworks and report critical safety incidents.

  • New York RAISE Act

    New York's Responsible AI Safety and Education Act, setting safety and transparency requirements for frontier model developers.

  • Illinois SB 315

    Illinois's frontier AI safety bill.

Frontier safety frameworks

Middle powers

Session 04

Topics Survey I

The first of two topic surveys covers international coordination and US–China competition, the governance of compute and chips, and AI's economic impacts and the policy responses under debate.

International coordination and China

Compute governance

Economics

Session 05

Topics Survey II

The second topic survey covers AI's risks to democratic institutions, cyber and CBRN threats, liability for AI harms, and the character of AI systems — including AI's growing role inside the executive branch.

Democracy

Cyber

CBRN

Liability

Character and “Exec AI”

Session 06

How to Have a Career in AI Policy

How can you contribute? This session covers policy skills, career pathways in Congress, think tanks, and the executive branch, networking, and public writing. Fellows then learn how to write a policy brief — executive summary, the problem, step-by-step recommendations, and why it matters — form groups, and begin drafting.

Public writing

  • Learning by Writing

    Holden Karnofsky on using writing as a tool for forming and testing your views.

Career resources

Writing a policy brief

  • Policy Briefs

    The UNC Writing Center's guide to the purpose, structure, and style of a policy brief.

  • Sample Policy Briefs

    Example policy briefs from Northwestern's Institute for Policy Research.

Session 07

Policy Brief Defense

Fellows conclude the program by presenting their policy briefs and defending them from peers' critique — a low-stakes session that encourages innovative, ambitious proposals. If you're stuck on a topic, the optional readings below are good starting points.

Datacenters

Robotics

RSI and AI progress

  • Measuring AI R&D AutomationOptional

    A paper on how to measure the extent to which AI research and development is being automated.

  • The Least Understood Driver of AI ProgressOptional

    Epoch AI on software progress — how improvements in algorithms and data quality reduce the compute needed for a given capability.

  • The Goodhart SingularityOptional

    Tom Reed argues that AI improving itself against internal benchmarks might only appear superintelligent, with gains that fail to generalize to the real world.

  • Taking Jaggedness SeriouslyOptional

    Helen Toner on the uneven, jagged profile of AI capabilities and what it means for policy.

Model weight security

Military