Skip to main content

Contributing

This is an open-source project and contributions are welcome. There are several ways to help.

Reporting errors​

If you find a factual error — a wrong settlement figure, an incorrect regulatory date, a misattributed incident — please raise a GitHub issue with:

  • The entry ID (e.g. C2, E1)
  • The specific claim that appears to be wrong
  • The correct information with a primary source link

Accuracy is the highest priority. Every verified correction improves the resource for everyone.

Suggesting new incidents​

Real-world incident examples make abstract risks concrete. If you know of a documented AI incident that belongs in an entry, raise an issue with:

  • The entry ID it relates to
  • A brief description of the incident
  • The date (approximate is fine)
  • A primary source link (news article, court record, official report)

Incidents must be real and documented. Illustrative examples are already flagged as such in entries.

Suggesting new entries​

The taxonomy currently covers 17 risk categories across 7 domains. If you believe a material AI risk is absent or inadequately covered, raise an issue describing:

  • The risk you believe is missing
  • Why it is distinct from existing entries
  • At least one documented real-world incident demonstrating the risk

New entries require editorial review before being added.

Improving controls content​

If you have practitioner experience implementing AI risk controls and believe the Layer 3 or Layer 4 content for an entry is incomplete, incorrect, or could be more actionable, pull requests are welcome.

Controls content must be:

  • Specific enough to implement
  • Referenced against a documented source or practice
  • Reviewed against the entry template schema

How to submit changes​

  1. Fork the repository on GitHub
  2. Create a branch for your change (git checkout -b fix/e1-saferent-figure)
  3. Make your changes following the entry template schema
  4. Submit a pull request with a clear description of what changed and why
  5. Reference any sources in the PR description

Entry template​

All entries follow the schema defined in schema/entry_template.md. Please follow the template when contributing new or updated content. Consistency across entries is a core quality requirement.

Standards​

  • All factual claims must be verifiable against primary sources
  • Uncertain claims should be flagged with [VERIFY] inline rather than silently included
  • Controls must be actionable — specific enough that a practitioner can implement them
  • Persona hooks must be written for the specific persona, not as generic risk language
  • No AI-generated content should be submitted without human review and verification

Code of conduct​

This project follows the Contributor Covenant code of conduct. Be respectful, constructive, and focused on improving the resource for practitioners.

Questions​

Raise a GitHub issue or start a GitHub Discussion for questions about the project, scope, or contribution process.