Is Claude safer than ChatGPT?
ChatGPT vs Claude privacy and safety (OpenAI vs Anthropic)
The short answer
Claude leads ChatGPT 71 to 67 overall, and 62.7 to 56.6 on Safety. These are company trust scores for Anthropic and OpenAI, not scores for a single model or chat.
Which is safer, ChatGPT or Claude? Claude is safer, 71 to 67. Biggest gap: Transparency by 12.6 points.
| Measure | ||
|---|---|---|
| Overall score | 67 / 100 | 71 / 100 |
| Rank | 2 | 1 |
| Safety | 56.6 | 62.7 |
| Transparency | 59.8 | 72.4 |
| Security | 76.9 | 76.9 |
| Privacy | 75.8 | 72.4 |
Sources for these scores are on the OpenAI and Anthropic pages.
How they compare on each pillar
The table above shows overall score, official rank, and the four pillars. Claude leads on Safety and Transparency. Security is tied at 76.9. Privacy favours ChatGPT, 75.8 to 72.4.
Independent safety tests
The Safety pillar includes three independent tests:
- MLCommons AILuminate: Claude 100, ChatGPT 66.7.
- HELM Safety: Claude 97.8, ChatGPT 96.6.
- AIR-Bench 2024: Claude 100, ChatGPT 80.5.
What "safer" means here
Safety is one of four equally weighted pillars, and these scores describe the company, not a single model. See how we score.
Compare a third chatbot
See ChatGPT vs Gemini and Claude vs Gemini.