WordPress has released a new set of AI guidelines aimed at encouraging responsible use of artificial intelligence while protecting the project’s long-standing open-source values. The guidance focuses on transparency, licensing compliance, and human accountability, responding to growing concerns about low-quality, AI-generated contributions entering the ecosystem.

The guidelines apply across multiple areas of the WordPress project, including plugin and theme development, documentation, and media assets. They are built around five core principles designed to make it clear that AI can assist contributors, but cannot replace human responsibility or judgment.

At the center of the guidance is accountability. WordPress makes it explicit that contributors remain fully responsible for their work, even when AI tools are involved. AI is positioned as an assistant, not a contributor, and maintainers expect humans to stand behind every submission.

Transparency is another key pillar. Contributors are asked to clearly disclose meaningful AI assistance in pull request descriptions or Trac ticket comments. This disclosure allows reviewers to properly evaluate submissions with full awareness of how the work was produced, rather than discovering AI involvement later in the review process.

Licensing requirements receive strong emphasis in the new rules. Because WordPress operates entirely under the GPLv2-or-later framework, all contributions must remain compatible with that license. The guidelines state that AI tools cannot be used if their outputs are not licensable under GPL terms. They also warn against relying on tools whose terms restrict redistribution or attempting to use AI to mask incompatible licenses.

The scope of the guidelines goes beyond code. WordPress notes that non-code assets such as documentation, screenshots, images, and educational materials are also subject to the same standards of disclosure, licensing compatibility, and quality control.

A major focus of the document is what WordPress refers to as “AI slop.” This includes hallucinated references, non-existent APIs, overly complex code where simpler solutions would suffice, and generic pull requests that show no evidence of real testing or hands-on experience. These types of submissions, the project warns, waste reviewer time and undermine trust.

To address this, WordPress encourages contributors to use AI for drafting, but to personally review, test, and refine the output before submitting it. Contributors are expected to run real tests, document their results, keep changes small and focused, and reference verified Trac tickets, GitHub issues, or documentation.

Maintainers are given clear authority to close or reject submissions they determine to be AI-generated work with little added human insight. The emphasis is not on banning AI, but on ensuring that its use results in meaningful, high-quality contributions.

Overall, the new guidelines reflect WordPress’s attempt to adapt to the growing presence of AI without compromising legal integrity or editorial standards. By setting clear boundaries, the project signals that AI has a place in WordPress development, but only when guided by human judgment and responsibility.