Bringing safe futures into view.

The behaviors that matter most for AI safety, like planning, patience, and deception, only show up over time. Horizon Research is an independent AI safety lab studying temporal cognition in language models: how they represent time, weigh near against far, and act over long horizons, and building evaluations that surface these behaviors in high-stakes environments.

Contact

Or write directly to shenk.justin@gmail.com

Get updates

Occasional research updates. No spam.