The Future of Life Institute's 2026 index grades labs on risk, transparency, and governance — Anthropic leads with a C+, while xAI, DeepSeek, and Mistral effectively fail.
The Future of Life Institute released its 2026 AI Safety Index. The highest grade was a C+, awarded to Anthropic; OpenAI and Google DeepMind landed at C, Meta at D+, and xAI, DeepSeek, and Mistral effectively failed. The index scores labs on risk management, transparency, governance, and whether they honour their own commitments.
No lab earning above a C is the headline. The frontier is shipping faster than its own safety practices can keep up.
Governance is becoming a scoreboard buyers and regulators will read. For operators choosing a model vendor, third-party safety grades are turning into procurement criteria — factor them in before you standardise.
The daily signal, curated. Get it in your inbox.
Subscribe on Substack →