An independent watchdog just handed the world's biggest AI companies their report cards — and the best grade anyone got was a C+.
The Future of Life Institute released its 2026 AI Safety Index on Tuesday, grading nine frontier AI labs across six domains including risk management, transparency, governance, and existential safety. The results paint a sobering picture of an industry that talks safety but rarely delivers it in practice.
The Rankings
- Anthropic: C+ (highest overall, leads 5 of 6 domains) - OpenAI: C (leads in Risk Assessment) - Google DeepMind: C - Meta: D+ (improved from 6th to 4th place) - Z.ai (formerly Zhipu AI): D - Alibaba Cloud: D - xAI: F (dropped from 4th to 7th place) - DeepSeek: F - Mistral: F (dead last — Europe's top AI company scores worst on safety)
Key Findings
**Safety rhetoric has outpaced actual behavior.** The report found that across Google DeepMind, OpenAI, and xAI, leadership's reassuring public messaging diverges sharply from commercial conduct and legislative stances — making stated commitments "an unreliable proxy for actual safety practice."
**Existential safety is the weakest domain across the entire industry.** No company scored above C-; most scored D or below. The panel judged current approaches — including interpretability and Chain-of-Thought monitoring — as "entirely inadequate," noting that "detection is not prevention."
**Several labs quietly walked back earlier safety promises.** Anthropic, OpenAI, Google DeepMind, and Meta have weakened or voided pledges to pause unilaterally if danger thresholds are approached, often citing competitor-contingent conditions. Reviewers called this a pattern of "moving goalposts" that has "undermined safety frameworks across the board."
**Military AI is accelerating with few guardrails.** The panel flagged the industry's pivot to military applications as an emerging risk. Companies including Anthropic, OpenAI, Google DeepMind, and Meta that previously banned military uses have gradually reversed course since 2024.
European dissonance: Despite the EU being a leader in AI safety regulation, Mistral — Europe's most prominent AI company — scored dead last.
The index was compiled by a seven-member independent panel including UC Berkeley's Prof. Stuart Russell, who warned: "Companies have backed away from earlier commitments to release new systems only with safety measures appropriate for their capability levels; now, they're planning to release them even if it's demonstrably unsafe to do so."
