A 10-week, in-person research training program for undergraduates and early-career researchers in AI safety and nuclear security.
An introductory survey course on existential and global catastrophic risk.
2024 SYLLABUS →A self-paced, comprehensive guide to understanding and defending against the next generation of AI threats.
From Fall 2025 to present
XLab announced Second Look, hosting its inaugural cohort of researchers for Summer 2026. Spearheaded by UChicago AI safety researcher Zephy Roe, the initiative focuses on replicating and reproducing popular AI safety studies, allowing for systematic verification of foundational AI safety claims and data transparency.
XLab is hosting its 5th summer research fellowship cohort at the University of Chicago, mid-June through late-August. Each fellow scopes their own research question, justifies its importance, identifies appropriate methods, and delivers a significant written product — essentially acting as their own “PI,” mentored by a domain expert.
XLab Director Daniel Holz joined a panel with Sasha Havlicek (CEO, Institute for Strategic Dialogue) and Will Rinehart (Senior Fellow, American Enterprise Institute) for a discussion on regulation, governance, and guardrails for AI. The discussion was moderated by technology and social media reporter Naomi Nix.
XLab’s AI safety group hosted Anthropic technical AI safety researcher Ross Nordby for a discussion on forecasting, AI timelines, and alignment, followed by audience Q&A on current AI safety research agendas and priorities.
XLab’s Fall 2025 Nuclear Winter and Fuel Loads working group produced a comprehensive report on the assumptions and uncertainties that exist in the literature on nuclear winter modeling, focused on how researchers can leverage modern open-source datasets and advanced climate modeling to advance the field’s understanding.
In collaboration with the Bulletin of Atomic Scientists, AMC Chicago River East hosted a screening of Kathryn Bigelow’s nuclear-war thriller, A House of Dynamite. The screening was followed by a discussion with Bulletin CEO Alex Bell and XLab Director Daniel Holz on the film’s central themes, including missile defense, decision-making, and launch authority.
XLab, along with the Bulletin of Atomic Scientists and the Norwegian Mission to the United Nations, hosted a side event for the UN’s annual First Committee meeting to discuss the resolutions produced from the Nobel Assembly in July 2025. The panel included XLab’s Daniel Holz, Karen Hallberg (Chair of Pugwash), Alex Bell (CEO of the Bulletin of Atomic Scientists), and Izumi Nakamitsu (UN Under-Secretary-General of Disarmament Affairs), with remarks from Physics Nobel Laureate (2011) Brian Schmidt.
XLab summer research fellows Zephy Roe, Julian Huang, and Jack Sanderson spent 10 weeks developing the XLab AI Security Guide: an open-source, self-paced and comprehensive guide to understanding and defending against the next generation of AI threats, preparing students to pursue robust AI security research. The guide provides source material for an upcoming XLab AI Security fellowship group during the academic year, building on the AI fundamentals course and the technical paper reading group.
During the Cuban Missile Crisis, John F. Kennedy estimated the risk of nuclear war to be "somewhere between one out of three and even."
Today, there are more vectors to potential catastrophe than ever before, and experts are increasingly warning that the chance of disaster this century is intolerably high — see Forecasting Existential Risks.
XLab is a research community focused on reducing the greatest threats to humanity. We actively seek collaborators and students from across the academy and beyond.
