WordPress's New AI Guidelines: What They Mean for Plugins, Themes and Contributions

WordPress has published official AI guidelines covering how AI can and can't be used when contributing plugins, themes, documentation and media assets to the project. The short version: AI is welcome, but you're still accountable for everything it produces, and low-effort output will be rejected.

Key takeaways

• WordPress explicitly encourages AI use, but contributors remain fully responsible for whatever AI helps them produce.
• You must disclose meaningful AI assistance in your pull request description or Trac ticket comment.
• Any AI-generated output must be compatible with GPLv2-or-later licensing. Tools whose terms block GPL use are off-limits.
• "AI slop" (hallucinated references, bloated code, untested PRs) can be closed or rejected by maintainers.
• Quality standards haven't changed. AI just changes how you get there, not the bar you have to clear.

The five principles WordPress has set out

The guidelines are built around five core principles, and they're worth reading word for word:

1. You are responsible for your contributions. AI can assist, but it isn't a contributor.
2. Disclose meaningful AI assistance in your PR description and/or Trac ticket comment.
3. Licence compatibility matters. Contributions must remain compatible with GPLv2-or-later, including AI-assisted output.
4. Non-code assets count too. Docs, screenshots, images and educational materials all fall under the same rules.
5. Quality over volume. Avoid low-signal, unverified "AI slop". Reviewers may close or reject work that doesn't meet the bar.

These aren't vague aspirations. Maintainers have been explicitly given the authority to act on them.

Why transparency matters here

The transparency requirement is practical, not symbolic. When a reviewer knows AI was involved and how it was used, they can evaluate the work more accurately. Hiding it doesn't help anyone, and if the output turns out to be hallucinated or untested, it wastes the reviewer's time and can introduce real problems into the codebase.

GPL licensing and tool choice

Licensing is at the heart of everything WordPress does. The platform runs under GPLv2, and everything built for it, including plugins and themes, must be open source under the same framework. That's non-negotiable, and the new AI guidelines extend that requirement to AI-generated output.

Specifically, the guidelines state two clear rules:

• Do not use tools whose terms forbid using their output in GPL-licensed projects or impose additional restrictions on redistribution.
• Do not rely on tools to "launder" incompatible licences. If an AI output reproduces non-free or incompatible code, it cannot be included.

In plain terms: if the AI tool you're using won't let you release its output under GPLv2, you can't use that tool for WordPress contributions. And you can't assume that running incompatible code through an AI somehow clears the licence issue. It doesn't.

This matters beyond the core project too. If you're building a plugin or theme for public distribution, the same GPL obligations apply. Our WordPress development service always builds with these licensing requirements in mind.

What counts as AI slop

The guidelines define AI slop specifically, which is useful because it's not just "bad AI writing". In the WordPress context, it means:

• Hallucinated references, such as links or APIs that don't actually exist.
• Overly complicated code when a simpler solution exists.
• GitHub PRs that are generic and don't reflect actual testing or real experience.

The recommended approach is straightforward: use AI to draft, then review it yourself properly. That means submitting small, concise PRs with clear commit messages, running and documenting real tests, and only linking to Trac tickets, GitHub issues or documentation you've actually verified.

Maintainers can close or reject contributions they judge to be AI slop with little added human insight. That's not a threat, it's just a recognition that reviewers' time is finite and the project's quality standards haven't shifted just because AI can generate text and code quickly.

What this means in practice

For anyone building on WordPress, whether that's contributing to the core project or shipping plugins and themes, these guidelines change the process rather than the destination. AI can genuinely speed up drafting, documentation and code generation. But a human still needs to read it, test it, understand it and stand behind it.

For businesses running WordPress sites, the guidelines are also a useful signal about the direction of the ecosystem. The WordPress project is actively trying to prevent a flood of low-quality, AI-generated plugins and documentation from degrading the platform. That's good news for anyone who relies on WordPress for their site's day-to-day reliability.

If you want to make sure your WordPress site is built and maintained to a proper standard, regardless of what tools are used to get there, take a look at our website maintenance service.

The bigger picture

These guidelines are fundamentally about preserving trust in the WordPress contribution process as AI becomes a normal part of development work. They don't discourage AI. They set clear boundaries around disclosure, licence compliance and quality so that the project stays legally sound and practically useful.

Requiring disclosure, enforcing GPL compatibility and giving maintainers the power to reject low-quality submissions protects both the legal integrity of the project and the time of everyone who reviews contributions. That's a reasonable position, and one that any serious developer working with AI tools should be able to get behind.