Safe, as far as we can see.

Horizon Research is an independent AI safety lab. We study temporal cognition in language models: how they encode time, weigh near against far, and behave over long horizons. We also build evaluations for AI in high-stakes environments.