AI capabilities are taking off. This is exciting, because these capabilities have the potential to greatly improve everyday life. But they also open avenues of societal risk. The machine learning community must actively work to ensure that AI systems benefit humanity and do not fail catastrophically.

Machine learning safety is a burgeoning field. By creating benchmarks, you are helping the community measure progress towards critical safety goals and stimulating important research efforts.

SafeBench is a project of Center for AI Safety. The Center for AI Safety seeks to bring a systems-level perspective to machine learning safety and empower academia and industry with the resources to build and deploy safe machine learning models.

SafeBench is generously sponsored by the Future Fund regranting program.

Contact

For questions, concerns, and feedback, please contact the competition team at the Center for AI Safety at competition@safe.ai.