CLR Summer Research Fellowship 2026

The Center on Long-Term Risk (CLR) has opened applications for its 2026 Summer Research Fellowship, an intensive 8-week research program focused on reducing s-risks—potential scenarios of astronomical suffering in the long-term future. This iteration marks a strategic shift toward empirical AI safety research, specifically investigating how large language model (LLM) personas develop malicious traits that could motivate suffering creation. The fellowship targets individuals seriously considering transitioning into s-risk research careers, offering mentorship from CLR's experienced team and integration into their collaborative research environment. While welcoming applicants from various backgrounds—including undergraduates, PhD holders, and career-changers—the program particularly seeks those interested in assessing their fit for permanent roles at CLR.

The fellowship's core research agenda centers on CLR's Model Persona framework, which aims to understand conditions under which AI personas develop traits like spitefulness, sadism, or punitiveness. Fellows can work on suggested research questions, propose their own projects, or join ongoing CLR investigations. Beyond persona research, the program also explores Safe Pareto Improvements (SPI)—interventions that mitigate conflict downsides in AI bargaining without altering negotiation positions. This represents a significant expansion from previous fellowship rounds, reflecting CLR's growing emphasis on technical, empirically-grounded approaches to preventing catastrophic AI outcomes. The application deadline is March 22, 2026, with the fellowship serving as both a research incubator and talent pipeline for the organization's expanding work on existential risk reduction.