Overview

FAR AI is a non-profit research institute focused on making advanced artificial intelligence safer and more beneficial for society. It conducts in-house research on AI safety topics such as model evaluation, interpretability, and robustness, and it also supports safety-driven research through collaborations and targeted grants. A key activity is red-teaming frontier AI systems to identify vulnerabilities and inform safety standards for developers and governments, while it also builds community through FAR.Labs in Berkeley and global dialogues like the International Dialogue on AI Safety. Its funding comes mainly from philanthropy, with profits from for-profit AI work capped at 10% to preserve independence, and its goal is to steer powerful AI toward minimizing risk and maximizing societal benefit through research, governance discussions, and broad collaboration.

Simplify's Take

What believers are saying

• $30M+ 2025 funding enables rapid research scaling.
• International dialogues shape global AI safety norms.
• Grantmaking influences major labs' safety directions.

What critics are saying

• Coefficient Giving withdraws funding by 2028 over breakthroughs.
• Anthropic poaches Berkeley researchers in 6-12 months.
• DWF model cuts org funding in 18-36 months.

What makes FAR AI unique

• FAR AI red-teams frontier models for OpenAI and EU CBRN risks.
• FAR.Labs Berkeley co-working fosters AI safety collaborations.
• FAR AI caps for-profit revenue at 10% for independence.

Help us improve and share your feedback! Did you find this helpful?

YesNo