This marks our first step to advance global testing standards for AI safety. The collaboration is a pioneering effort to develop safety benchmarks for generative AI.
By leveraging the expertise and resources of Foundation and MLCommons, we aim to develop robust benchmarks that will set a new standard for AI safety. Both organisations will contribute towards the Safety Benchmarks, collaborate to engage our network of partners and communities, and promote the Safety Benchmarks globally.
You can use Project Moonshot to access and test v0.5 of the Safety Benchmark as well as contribute benchmarks.
