Skip to content

AI Attribution Governance: Enforcing AI Disclosure Policies at the CI Level

The open-source ecosystem is converging on a hard question: when a commit is written with AI assistance, how do we know — and how do we enforce the disclosure policy?

Python's discourse, Linux kernel's Assisted-by trailer, Fedora's AI policy, Apache's disclosure guidelines — every major project is grappling with this. But until now, there has been no tool at the CI level to enforce whatever policy a project chooses.

Commit Check v2.11.0 introduces AI Attribution Governance — a new feature that detects known AI tool signatures in commit messages and lets projects decide whether to forbid them outright. To our knowledge, no existing tool enforces this kind of policy at the CI level.

The industry need

The conversation around AI disclosure is no longer theoretical:

  • The Linux kernel standardized on the Assisted-by: trailer format — but deliberately stopped short of CI enforcement. As Sasha Levin noted at the Maintainers Summit, the kernel sets the convention, not the gate.
  • The Python community is actively discussing whether Claude Code usage should be documented
  • VS Code issue #313962 proposes replacing Co-authored-by with Assisted-by for AI agents
  • Fedora requires AI disclosure (recommends the Assisted-by trailer). QEMU and Gentoo go further and forbid AI-generated contributions entirely.

Each community defines its own policy — but none provides a neutral enforcement layer. That is the gap Commit Check fills.

Configuration: a single toggle

Commit Check keeps it simple. One configuration value, three ways to set it:

[commit]
ai_attribution = "forbid"

commit-check --message --ai-attribution=forbid

CCHK_AI_ATTRIBUTION=forbid commit-check --message

Two modes:

Mode Behavior
"ignore" No validation (default, backward compatible)
"forbid" Rejects any commit containing known AI tool signatures

There is no require mode in this release — only ignore and forbid. The reason is pragmatic: requiring a Assisted-by or similar trailer is a substantially harder problem (validating semantics, not just pattern-matching), and the most immediate demand from projects is the ability to say no. The kernel and Fedora communities that want require are on the roadmap (see What's next).

Detected AI tool signatures

Commit Check ships with a curated database of known AI tool markers. The detection covers multiple signature formats per tool — Co-authored-by, Assisted-by, body markers, and model names:

AI Tool What gets detected
Claude Code Co-authored-by: Claude, Assisted-by: Claude:<model>, emoji markers, Claude-Session:, Claude-Workflow:
GitHub Copilot Co-authored-by: Copilot
OpenAI Codex Co-authored-by: Codex
Gemini Co-authored-by: Gemini
Cursor Co-authored-by: Cursor
Devin Co-authored-by: Devin
Aider Co-authored-by: Aider, Co-authored-by: ... (aider)
Windsurf Co-authored-by: Windsurf
Tabby Co-authored-by: Tabby
Generic AI Assisted-by: <tool>:<model> [tools], model names like claude-sonnet-4, gpt-4-turbo

Built-in false positive prevention

A Co-authored-by: Claude could theoretically be a human named Claude — but in practice, AI tools use known noreply email addresses. Commit Check anchors its detection to these, so:

🚫 Co-authored-by: Claude <noreply@anthropic.com> — detected
🚫 Assisted-by: Claude:claude-sonnet-4-20250514 [tools] — detected
✅ Co-authored-by: Claude Monet <monet@impressionism.fr> — ignored
✅ Co-authored-by: Jane Doe <jane@example.com> — ignored

The kernel-style Assisted-by: format also handles optional trailing tool lists correctly:

Assisted-by: Claude:claude-sonnet-4-20250514 coccinelle sparse

Only the AI tool marker is matched — the tool list is preserved as-is.

See it in action

With a config file containing ai_attribution = "forbid":

# This commit message would be REJECTED
echo "fix: resolve race condition

Co-authored-by: Claude <noreply@anthropic.com>" | commit-check -m

[FAIL] ai-attribution: Commit message contains known AI tool signature: Claude

# This commit message passes cleanly
echo "fix: resolve race condition

Co-authored-by: Jane Doe <jane@example.com>" | commit-check -m

[PASS] commit message is valid

Integration across the ecosystem

The feature is available across nearly every surface of Commit Check:

  • CLI: --ai-attribution=forbid
  • TOML config: [commit] ai_attribution = "forbid"
  • Environment variables: CCHK_AI_ATTRIBUTION=forbid
  • Python API: validate_message() returns AI attribution results
  • --format json: AI check status included in structured output
  • MCP Server (commit-check-mcp): synced in v0.1.7
  • GitHub Action (commit-check-action): coming in the next release

Scope and limitations

AI Attribution Governance detects the default behavior of AI coding tools — the trailers, markers, and metadata they add automatically. It is not designed to catch intentional circumvention. If a developer manually removes the AI signature before committing, this feature will not flag it.

This is the same trust boundary that every linter operates within: --no-verify bypasses pre-commit hooks, and a determined author can always rewrite history. The goal is to set a visible, enforceable policy for the standard case — making AI disclosure the path of least resistance — and leave intentional evasion to code review and engineering culture.

What's next

AI attribution governance in v2.11.0 is the foundation. Future work includes:

  1. require mode — reject commits that are missing an Assisted-by trailer, serving projects like the Linux kernel and Fedora that mandate disclosure
  2. PR summaries — show AI disclosure status per commit in pull requests
  3. MCP improvements — AI agents query describe_validation_rules to auto-comply before writing a commit
  4. Richer JSON metadata — structured AI signature data for SBOM and audit tooling

Try it today

pip install commit-check==2.11.0
echo "feat: add streaming support" | commit-check -m --ai-attribution=forbid

Or add it to your cchk.toml:

[commit]
ai_attribution = "forbid"

And let CI enforce your AI disclosure policy — automatically, on every commit.