Tabletop Exercise Lead, The Anthropic Institute

Anthropic•San Francisco, CA

10d•Hybrid

About The Position

The Tabletop Exercise Lead will build and lead The Anthropic Institute’s tabletop exercise program and the team that runs it. In consultation with technical experts and based on emerging trends, the Lead will set the research agenda: determining which questions exercises should answer, which scenarios and audiences to use, and what each exercise is intended to test. The Lead will own methods that make results comparable across runs, turn results into findings usable outside the exercise room, hire and manage a team and facilitator bench, and serve as the program’s public voice. Tabletop exercises give people who may face difficult decisions under uncertainty an opportunity to practice before those situations occur. They can reveal what tools and information participants reach for, what they lack, and how people and institutions might respond to unprecedented situations. Repeating scenarios with different groups—and with AI agents playing in simulations—can help generate empirical evidence about the distribution of decisions and possible outcomes. The role includes designing exercises as empirical studies, developing measurement and analysis practices, publishing findings, briefing senior audiences, maintaining a refreshed library of tested scenarios, creating a facilitator kit for internal teams and external partners, and directing tools such as Claude-driven simulations that generate a live outside-world feed during play. Anthropic’s location-based hybrid policy currently expects staff to be in an office at least 25% of the time, though some roles may require more. The position requires a bachelor’s degree or equivalent combination of education, training, and/or experience, in a relevant field demonstrated through coursework, training, or professional experience.

Requirements

Have designed and run tabletop exercises, wargames, or simulations for senior decision-makers and published on the subject.
Have training in a discipline that uses simulation, experimental methods, or structured analysis.
Have 10+ years of experience in national security, government, technology policy, or strategy.
Keep scenarios technically grounded with a strong grasp of AI capabilities and safety research.
Have managed a team, including hiring, setting direction, giving feedback, and developing people.
Use Claude and other AI tools extensively in own work.
Care about AI safety and Anthropic’s public benefit mission.
Minimum education: bachelor’s degree or an equivalent combination of education, training, and/or experience.
Required field of study: a field relevant to the role as demonstrated through coursework, training, or professional experience.

Nice To Haves

Published research on wargaming methods, simulation, or crisis decision-making.
Run exercises that fed into real policy or crisis-response decisions.
Run multi-iteration or experimental wargames and analyzed results across runs.
Used AI agents or language models as players, adjudicators, or analysts in simulations.
Led a research program or team through a period of growth.
Existing relationships across national security, government, policy, and AI, and trusted standing in those communities.
Published widely read work on AI, emerging technology, strategy, or crisis decision-making.

Responsibilities

Set the program’s research agenda, including the questions exercises should answer, scenarios to run, and audiences for each.
Design exercises as empirical studies by defining what each run tests, standardizing scenarios and data capture, and using repeated play to map the range of outcomes.
Build measurement practices, including pre- and post-exercise assessments, decision logs, transcript analysis, and after-action reports.
Publish findings from exercises and brief senior people inside and outside Anthropic on the gaps they expose.
Hire, manage, and develop a team and a bench of facilitators.
Maintain a library of tested scenarios for reuse and refresh it regularly so scenarios stay timely.
Build a facilitator kit (“TTX in a box”) so internal teams and external partners can run exercises to the same standard.
Create and direct exercise tools, including Claude-driven simulations that generate a live outside-world feed during play.
Serve as the program’s public voice.