# AI-Assisted Contributions Policy ## Purpose and Scope Godot Launcher permits the responsible use of AI-assisted tools by maintainers and external contributors. This policy applies to code, tests, documentation, translations, and any other material submitted to the project. ## Core Policy AI-assisted tools may be used during development, but their output is treated as untrusted input. Contributors are responsible for understanding, reviewing, adapting, testing, and maintaining everything they submit. Contributions that are unverified, cannot be explained by their author, or create licensing, privacy, or security concerns will not be accepted. Using an AI-assisted tool does not transfer responsibility for a contribution to the tool or its provider. The contributor remains accountable for the accuracy, behavior, quality, and long-term maintainability of the submitted work. ## Contributor Responsibilities Before submitting AI-assisted work, contributors must: - Understand the change and be able to explain its behavior, design, and relevant trade-offs. - Review and adapt generated material so it follows the project's conventions and addresses the intended problem. - Verify factual claims and test behavior to a level appropriate for the change. - Ensure they have the right to submit the material and preserve any required license notices or attribution. - Protect credentials, personal data, private security reports, confidential information, and other sensitive material. - Respond meaningfully to review feedback and revise the contribution when necessary. ## Review and Acceptance AI-assisted contributions are reviewed under the same quality, security, licensing, and maintainability standards as other contributions. Maintainers may ask for explanations, supporting evidence, tests, or revisions needed to evaluate a submission. Maintainers may reject work when the contributor cannot demonstrate an understanding of it, its behavior has not been adequately verified, it cannot be reasonably reviewed or maintained, or licensing, privacy, security, or quality concerns remain unresolved.