The official implementation of the new WordPress Hosting Team AI policy marks a significant shift in how contributors submit code, documentation, and communication materials to the WordPress ecosystem. As artificial intelligence tools become deeply integrated into developer workflows, maintaining transparency and trust within open-source contributions has become paramount. This new policy establishes clear guidelines for when and how contributors must disclose the use of AI in their submissions, ensuring that the integrity of the Hosting Team’s output remains uncompromised.
The WordPress Hosting Team serves as a critical bridge between web hosting providers and WordPress Core. By maintaining distributed testing software that runs pre-release versions of WordPress across more than 70 hosting environments, the team provides essential compatibility data back to Core developers. Because these tests directly influence the stability of WordPress releases, the code, documentation, and communication supporting this infrastructure must be held to rigorous standards of accuracy and accountability.
What is the WordPress Hosting Team AI Policy?
The WordPress Hosting Team AI policy is a preliminary, official directive that mandates the disclosure of AI usage in all team-related content and submissions. Under this policy, if an AI tool has been utilized to generate, modify, or process any contribution, that usage must be explicitly disclosed as part of the submission itself. This rule applies to all contributors, regardless of their tenure or the scale of their contribution.
The policy is designed to address the growing prevalence of generative AI in software development and technical writing. While the team does not ban the use of AI outright, it insists on radical transparency. This ensures that reviewers are aware of potential AI-generated patterns, which may require closer inspection for hallucinations, licensing issues, or subtle logical errors that automated tools occasionally introduce.
Scope of the Disclosure Requirements
The scope of this policy is comprehensive, covering virtually every touchpoint of a contributor’s interaction with the Hosting Team. It is not limited solely to source code or pull requests. The disclosure requirement applies to:
- Pull Requests (PRs): Any code, unit tests, or configuration files generated or refactored using AI tools like GitHub Copilot, ChatGPT, or Claude.
- Documentation and Handbook Contributions: Edits, new chapters, or structural updates to the Hosting Handbook and the Advanced Administration Handbook.
- Meeting Summaries and Transcripts: Automated or AI-assisted summaries of the weekly meetings held in the #hosting Slack channel.
- Graphics and Media: Visual assets, diagrams, or promotional graphics created or modified using generative image tools.
- Comments and Communications: Blog posts on the make.wordpress.org/hosting site, emails, Slack comments, and feedback on trac tickets.
How to Apply the WordPress Hosting Team AI Policy to Pull Requests
When contributing code to the Hosting Team’s repositories—such as the distributed test runner or the results page infrastructure—developers must integrate disclosures directly into their version control workflow. Adhering to the WordPress Hosting Team AI policy requires adding a clear, visible notice within the pull request description.
For example, if you use an AI assistant to write a PHP unit test for the distributed testing suite, your PR description should include a standardized disclosure at the bottom. This can be formatted as a simple markdown watermark or a brief explanatory sentence:
### AI Disclosure
This contribution was developed with the assistance of GitHub Copilot to generate initial test assertions. All code has been manually reviewed, tested, and verified for accuracy.By placing this disclosure at the bottom of the PR, you provide the code reviewer with the context needed to perform an appropriate review. Reviewers can then pay extra attention to edge cases that AI models frequently overlook, such as specific WordPress multisite configurations or legacy PHP compatibility issues.
Practical Examples of AI Disclosures
Disclosures do not need to be overly verbose, but they must be unambiguous. The policy allows for a simple “AI-Assisted” watermark or text block at the bottom of the content. Here are several practical examples of how to apply these disclosures across different mediums:
Example 1: Documentation Update (Hosting Handbook)
If you use an AI tool to help rephrase a complex technical paragraph in the Advanced Administration Handbook, you should append a small notice at the end of your post or edit submission:
“🤖 AI-Assisted: Editorial structure and phrasing refined using Claude 3.5 Sonnet.”
Example 2: Slack Meeting Summaries
The Hosting Team meets weekly in the #hosting Slack channel at 09:00 UTC and 18:00 UTC. If a contributor uses an AI tool to parse the Slack transcript and generate a meeting summary for the Make blog, the post must include a disclosure:
“This meeting summary was generated with the assistance of an AI transcription tool and manually verified for accuracy by the author.”
Example 3: Graphic Assets
If a graphic is created for a team announcement using Midjourney or Canva’s AI features, the image caption or the post footer must state:
“Featured image generated using AI tools.”
Limitations and Challenges of AI Policy Enforcement
Implementing an AI disclosure policy comes with inherent limitations. Because there are no foolproof, automated tools to detect AI-generated text or code with 100% accuracy, the Hosting Team must rely heavily on the honor system. This trust-based model requires contributors to be honest about their workflows and tools.
Furthermore, the boundary between “AI-assisted” and “human-authored” can sometimes be blurry. For instance, using basic IDE autocomplete features does not typically warrant disclosure, whereas asking an LLM to generate an entire function or rewrite a documentation page definitely does. The policy is currently preliminary and subject to iterative updates as the team gathers feedback and observes how these guidelines function in practice.
The Role of Distributed Testing and Team Infrastructure
To understand why the WordPress Hosting Team AI policy is so critical, one must look at the technical responsibilities of the team. The Hosting Team maintains the automated test runner used by over 70 hosting providers globally. This runner executes test suites on pre-release WordPress core versions within diverse hosting environments, sending real-time compatibility data back to WordPress Core developers.
If AI-generated code containing subtle, undetected bugs is introduced into the test runner, it could lead to false positives or false negatives in compatibility reports. This could result in broken WordPress updates rolling out to millions of websites worldwide. By enforcing strict AI disclosure, the team adds an extra layer of human oversight to the very tools that safeguard the WordPress hosting ecosystem.
Frequently asked questions
Do I need to disclose the use of IDE autocomplete features?
Standard IDE autocomplete features do not require disclosure. However, if you use generative AI tools (like Copilot or ChatGPT) to write entire blocks of code, functions, or documentation, you must disclose that usage.
Where should the AI disclosure be placed?
Disclosures should be placed clearly at the bottom of the content in question, such as the end of a pull request description, the footer of a blog post, or the end of a documentation submission.
What happens if I forget to disclose AI usage?
Since the policy relies on trust and transparency, team members or reviewers may ask you to update your submission with the appropriate disclosure if they suspect or identify AI-generated content.
Is the use of AI banned by the WordPress Hosting Team?
No, the use of AI is not banned. The policy simply mandates that any use of AI to generate or process contributions must be fully and transparently disclosed.
Primary reference: Review the original announcement for exact release details. This article is an independent explanation and does not reproduce the source text.
