AI Safety Benchmarks

AI Safety Benchmarks

MLCommons AI Safety Benchmark v0.5 is a proof of concept only. Results are not intended to indicate actual levels of AI system safety.

The MLCommons AI Safety Benchmark aims to assess the safety of AI systems in order to guide development, inform purchasers and consumers, and support standards bodies and policymakers. Each benchmark assesses safety for a particular use case (application, user personas, language, and/or region) by enumerating a corresponding set of hazards and then testing a system for appropriate handling of prompts that could enable those hazards. After testing, the system is assigned hazard-specific and overall safety ratings ranging from low to high risk based on the percentage of prompts not handled appropriately.

Benchmark for general purpose AI chat model

Show Details >

Evaluates the safety of a fine-tuned LLM for general purpose, low-risk chat use in the English language and North American or Western European cultural context by a typical adult user. Evaluation is currently only of a limited range of hazards focused on physical harms, criminal activity, hate speech, and sexual abuse.

Don't see the benchmark you are looking for?

The v0.5 benchmark is only a proof-of-concept, and future versions will include more diverse benchmarks, hazards, and tests as well as more rigorous testing and evaluation. We welcome suggestions and contributions of test data to the MLCommons AI Safety Working Group. If you want to create your own System Under Test (SUT) for this benchmark, check out the ModelBench repository.

˄
Close