Home / Labs / Center for AI Safety
#143 on the AGIX Leaderboard · score 0
Center for AI Safety
Research and field-building; co-created Humanity's Last Exam.
| Type | Safety & alignment |
|---|---|
| Founded | 2022 |
| Founders | Dan Hendrycks |
| Headquarters | San Francisco, US |
| Website | safe.ai |
Center for AI Safety models
0models released, last 12 months
8public models in total
Latest releases
- cais/Yi-34B-Chat_RMU17 Apr 2024
- cais/Mixtral-8x7B-Instruct_RMU16 Apr 2024
- cais/Zephyr_RMU16 Apr 2024
- cais/HarmBench-Mistral-7b-val-cls17 Mar 2024
- cais/zephyr-7b-beta-CUT12 Mar 2024
- cais/HarmBench-Llama-2-13b-cls-multimodal-behaviors27 Feb 2024
Most downloaded
- cais/HarmBench-Llama-2-13b-cls36,262 downloads (30 days)
- cais/HarmBench-Mistral-7b-val-cls3,112 downloads (30 days)
- cais/HarmBench-Llama-2-13b-cls-multimodal-behaviors787 downloads (30 days)
- cais/Zephyr_RMU469 downloads (30 days)
- cais/zephyr_7b_r2d272 downloads (30 days)
Source: Hugging Face Hub, organisation cais, read 2 Nov 2026.
Center for AI Safety: quick answers
Who founded Center for AI Safety?
Center for AI Safety was founded in 2022 by Dan Hendrycks.
People at Center for AI Safety
Similar labs
#74GoodfireSan Francisco, US · 2024
Interpretability research and tools (Ember).
7 research roles#82Apollo ResearchLondon, UK · 2023Evaluations for deceptive behaviour in AI models.
5 research roles#103METRBerkeley, US · 2022Evaluates frontier models for dangerous autonomous capabilities.
2 research roles#115Redwood ResearchBerkeley, US · 2021Non-profit research on AI control and alignment.
1 research roles#116Epoch AIUS · 2022Research on AI trends: compute, data and benchmarks.
1 research roles#117UK AI Security InstituteLondon, UK · 2023UK government body testing advanced AI models.
1 research rolesSomething wrong or missing? Send a correction. We fix sourced errors within a day.