About 200 people work on safety at OpenAI.

TL;DR

Approximately 200 people are employed in safety roles at OpenAI.
OpenAI's safety efforts are structured across multiple teams: safety systems, preparedness, alignment, and model policy.
The safety systems team focuses on guardrails and evaluations.
The preparedness team manages OpenAI's preparedness framework.
The alignment team concentrates on training models to align with human values.
The model policy team is responsible for developing the model specification.
Research from dissolved teams, such as the superalignment and AGI readiness teams, is being absorbed by other existing teams.

Continuer la lecture de l'article original