towardsafe.si · toward safe superintelligence
The most valuable idea in AI right now is safety.
The labs below aren't building chatbots. They're working on alignment, evaluations, and provable safety, and investors are pouring billions in. This domain is for whoever wins that race.
-
AN
Anthropic
$380B
Constitutional AI. The safety-first frontier lab, founded by ex-OpenAI researchers.
anthropic.com ↗ -
SS
Safe Superintelligence
$32B
Sutskever's lab. One goal, one product: safe superintelligence. No product shipped, by design.
ssi.inc ↗ -
ME
METR
$71M
Independent safety evaluations of frontier models. Refuses funding from AI labs.
metr.org ↗ -
FA
FAR.AI
$30M+
Technical AI safety research, funded by Coefficient Giving and Schmidt Sciences.
far.ai ↗ -
CO
Conjecture
Safety
AI safety research lab working on scalable alignment for transformative AI.
conjecture.dev ↗ -
RR
Redwood Research
Safety
Nonprofit tackling AI alignment, from interpretability to adversarial robustness.
redwoodresearch.org ↗
Figures from widely reported funding rounds, 2024–2026. Cards link to each lab.
