The UNLOCK call was launched by the Helmholtz Association to promote benchmarking as a foundation for trustworthy and reproducible AI in science. Coordinated within the Helmholtz Information and Data Science Framework, UNLOCK supports the development of high-quality, open, and multimodal benchmarking datasets across domains such as health, climate, energy, and materials science. The selected projects initiate a Helmholtz-wide effort to establish AI benchmarking as a cornerstone of comparable and trustworthy AI, laying the foundation for future initiatives that will further expand this field across the Association. These projects lay the groundwork for a Helmholtz-wide AI benchmarking ecosystem.

The initiative connects domain scientists and AI experts to establish shared standards and reproducible pipelines for AI benchmarking. They benefit from close collaboration with Helmholtz AI, Helmholtz Imaging, HIFIS and HMC, and contribute to building a Helmholtz-wide community of practice for AI benchmarking.

Each project receives funding from the Helmholtz Initiative and Networking Fund and will publish its datasets openly to foster collaboration and advance AI-driven scientific discovery.