The new WordPress AI Guidelines just landed in the official handbook, and if you’re a contributor or a dev planning to push code to Core, you need to pay attention. We aren’t in the “wild west” era of ChatGPT copy-pasting anymore; the core team is setting clear expectations for transparency, licensing, and quality.
These guidelines aren’t here to ban your favorite LLM. Consequently, they focus on responsible usage and professional accountability. I’ve spent enough time debugging “hallucinated hooks” to know why these rules are necessary. Furthermore, if you’re interested in the deeper debate on quality, check out recent discussions on why code quality still trumps speed.
Breaking Down the WordPress AI Guidelines
At a high level, the guide focuses on five core principles that every senior developer should already be practicing, but now they are codified. Specifically, the guidelines focus on:
- Individual Responsibility: You are responsible for your contributions. AI can assist, but it is not a contributor. If the code breaks, it’s on you, not the model.
- Meaningful Disclosure: Transparency is key. You must disclose meaningful AI assistance in your PR descriptions or Trac ticket comments.
- License Compatibility: All contributions must remain compatible with GPLv2-or-later. Therefore, you must ensure your AI tool’s terms don’t forbid this.
- Holistic Guidelines: This doesn’t just cover PHP or JS. It includes documentation, screenshots, images, and educational materials.
- Quality Over Volume: Avoid low-signal “AI slop.” Reviewers have been given the green light to reject work that doesn’t meet the project’s quality bar.
The GPL and “Vibe Coding” Risk
Licensing is where things often get messy in the WordPress ecosystem. The requirement for GPL compatibility is non-negotiable. If you’re using a tool that claims ownership of the output or restricts its redistribution, you can’t use it for WordPress Core. Consequently, developers must be more vigilant about the provenance of the code they ship.
I’ve seen many devs fall into the trap of “vibe coding,” where they trust the model’s architecture blindly because the syntax looks correct. However, outsourcing your logic to an AI usually leads to massive technical debt and “race conditions” that are a nightmare to debug later.
Final Takeaway for Contributors
The space is moving fast, and these guidelines are intended to be living documentation. You should read the official AI Guidelines handbook entry and share your feedback. Specifically, the team is looking for input from maintainers who are already handling AI-assisted contributions at scale. Don’t be the dev shipping unverified slop; be the one shipping standard-compliant, disclosed, and performant code.
