Safety Research Organizations

Summary

These organizations focus primarily on AI safety research rather than building commercial AI products. They include technical research labs, evaluation organizations, governance research groups, and academic centers.

Technical Research

Technical Safety Research

research

MIRI

Machine Intelligence Research Institute. Focuses on agent foundations and alignment theory.

research

ARC

Alignment Research Center. Works on eliciting latent knowledge and alignment evals.

research

Redwood Research

Focuses on interpretability and adversarial robustness.

research

Conjecture

Works on interpretability, cognitive emulation, and AI control.

research

FAR AI

Focuses on robustness, adversarial testing, and AI security.

Evaluations & Auditing

Evaluation Organizations

research

METR

Model Evaluation and Threat Research. Dangerous capability evaluations.

research

Apollo Research

Focuses on detecting deceptive AI behavior and scheming.

Policy & Governance Research

Governance Research

research

GovAI

Centre for the Governance of AI. AI governance research and policy.

research

CAIS

Center for AI Safety. Advocacy, field-building, and safety research.

Academic

Academic Groups

academic

CHAI

Center for Human-Compatible AI at UC Berkeley.

research

Epoch AI

Research on AI trends, compute, and forecasting.