Join the DARR community.
Maintained and grown by its community, with full attribution, by name, for every contribution. Two ways to take part:
Join with the form
Tell us what brings you here. A maintainer replies by email.
Contribute directly on GitHub
Open an issue or a pull request. Merged work carries your name. Ready-made templates for:
Found DARR useful? A star on the repository helps others discover it and grows the community.
Star on GitHub ↗Contributions are licensed CC BY 4.0.
Where the community is heading
The plan, in three steps.
More contributors
Bring in more people to help maintain and grow the register.
More frameworks
Map DARR to more standards, released on a regular cycle.
Community review
Each mapping checked by the community that maintains the source framework.
Ideas ready to pick up
Concrete starting points, sources linked. Bring your own too.
Extend the framework crosswalk
Map the register to more standards and control frameworks, so a DARR risk traces straight into the controls a team already runs. First up: the Cloud Security Alliance AI Controls Matrix (AICM), 247 control objectives across 18 domains. Good follow-ons include further ISO/IEC standards, sector regulations, and national AI frameworks.
Automate freshness with CI/CD
Every source the register draws on is versioned and public. Build pipelines that watch these upstreams and open a pull request when one changes, so the crosswalk never falls behind the frameworks it maps.
Make DARR a common language for developers
Help AI tool builders and application teams reference DARR risk ids (MR-001 to MR-082) as a shared vocabulary, in scanners, evaluations, model cards, and governance tooling. Linters, plugins, SDK mappings, and any integration that emits or consumes DARR ids are welcome.
Grow the community
Connect the register with the groups working the same problem, so each mapping is reviewed by the people who maintain the source it points to. Know a community that should be involved? Open an issue and say so.
Improve the register itself
Corrections, new sub-risks, sharper descriptions, and additional cross-checks. Every risk has a stable id and a permanent page; propose changes against the open dataset.