AI code attribution (provenance)
Recording which code was written by an AI agent, under whose direction and with which tool, in commits, pull requests and audit logs.
The common mechanism is a commit trailer: coding agents add a line such as Co-Authored-By naming the model, using the same Git trailer format GitHub uses for commits with multiple authors. Larger setups add a machine-written audit trail per run: the prompt or spec, the model, the tools called and the checks that passed.
Customers and auditors ask about it in security questionnaires and SOC 2 or ISO 27001 reviews, and lawyers ask who owns the result (source code ownership). Attribution also feeds metrics such as first-pass success rate. The person who directed and approved the change stays accountable; attribution records how it was made, not who is responsible.
Sources
- Git: git-interpret-trailers Documentation git-scm.com
- GitHub Docs: Creating a commit with multiple authors docs.github.com
Related terms
- Source code ownership (IP assignment): Who legally owns the code you paid for. In Poland, a contractor’s code becomes yours only through a written transfer of rights.
- SOC 2 and ISO 27001: The two security attestations enterprise buyers ask vendors for: an auditor’s SOC 2 report and an ISO/IEC 27001 certificate.
- Coding agent: An agent that reads a repository, edits files, runs commands and tests, and opens pull requests.
- AI acceptable use policy: A short company rulebook for staff: which AI tools are approved, what data may go into them, how output is checked and when AI use must be disclosed.
- First-pass success rate: The share of tasks a coding agent completes in one run that pass checks and review without a person rewriting them; the main health metric of agentic delivery.
