AI Provider Scorecard
Evaluate and compare AI vendors against your organization's values and technical requirements. Adjust weights by mission type — the ranking shifts accordingly.
Select your organization type to apply recommended weights, or adjust the sliders manually. The composite score reflects your priorities — not a universal ranking.
Organization type
Military & surveillance25%
Privacy & data25%
Labor & supply chain25%
Safety & harm response25%
Provider scores
| Provider | Morality | Capability | Score | Tier |
|---|---|---|---|---|
Anthropic (Claude) US · Private | 7.5 | 8.6 | 7.9 | Recommended |
Mistral France · GDPR-governed | 7.5 | 5.5 | 6.7 | With caution |
Google (Gemini) US · Public | 4.8 | 8.7 | 6.3 | With caution |
OpenAI (ChatGPT) US · Private | 4.0 | 8.6 | 5.8 | With caution |
DeepSeek China · State-adjacent | 3.3 | 7.8 | 5.1 | With caution |
xAI (Grok) US · Private | 2.6 | 7.3 | 4.5 | Not recommended |
Meta AI (Llama) US · Public | 3.4 | 5.6 | 4.3 | Not recommended |
Morality = weighted average across 4 dimensions. Capability = average of SWE-bench and GPQA benchmarks. Composite = 60% morality + 40% capability. Scores as of early 2026.
Methodology: read the full article →