Jason Adams @jason_the_adams (Team RepTeam RepA Team Rep is a person who represents the Make WordPress team to the rest of the project, make sure issues are raised and addressed as needed, and coordinates cross-team efforts.) – Automattic
AI 1.5.0 has been released and is available for download! “What’s new in AI…” posts (labeled with the #ai-release tag) are posted following every AI pluginPluginA plugin is a piece of software containing a group of functions that can be added to a WordPress website. They can extend functionality or add new features to your WordPress websites. WordPress plugins are written in the PHP programming language and integrate seamlessly with WordPress. These can be free in the WordPress.org Plugin Directory https://wordpress.org/plugins/ or can be cost-based plugin from a third-party. release, showcasing new features included in each release.
We’re pleased to announce the release of AI 1.5.0, the latest update to the canonical plugin powering AI-powered features in WordPress. This release introduces six new Experiments, including Text to Speech, Content Verification, MCP Access, WebMCP, Revision Change Summary, and BlockBlockBlock is the abstract term used to describe units of markup that, composed together, form the content or layout of a webpage using the WordPress editor. The idea combines concepts of what in the past may have achieved with shortcodes, custom HTML, and embed discovery into a single consistent API and user experience. Container Label Generation. It also expands the Embeddings infrastructure, improves Content Translation and Markdown Feeds, introduces additional AI providerProviderAn AI service offering models for generation, embeddings, or other capabilities (e.g., Anthropic, Google, OpenAI). connectors, and brings several Abilities into alignment with WordPress coreCoreCore is the set of software required to run WordPress. The Core Development Team builds WordPress..
What’s new in 1.5.0?
New Experiment: Text to Speech
The new Text to Speech experiment lets editors generate an audio version of a post, giving visitors the option to listen to content instead of reading it.
When enabled, editors can generate audio from a post’s title and content using a configured AI provider that supports speech generation. Audio processing runs in the background, allowing editors to continue working or return later while longer posts are processed.
Once generated, the audio file is saved to the Media Library and can be played directly in the editor. Editors can choose whether to display an audio player on the published post and regenerate the audio when needed. This provides another way for publishers to make their content accessible and available in different formats without requiring a separate audio production workflow.
New Experiment: Content Verification
The new Content Verification experiment introduces support for detecting C2PA Content Credentials in images uploaded to WordPress.
Content Credentials are a standard for attaching provenance information to digital media, helping publishers and audiences understand where content originated and how it may have been modified. When enabled, the experiment detects embedded C2PA data in uploaded JPEG, PNG, and WebP images and preserves the manifest information alongside the attachment. Credential status is displayed in the Media Library, including a new indicator in the grid view.
This initial implementation is read-only. It detects and preserves existing Content Credentials but does not yet perform cryptographic verification or generate new credentials for images. The experiment establishes a foundation for future content provenance features, including deeper verification and support for preserving credentials through WordPress image-processing workflows.
New Experiment: MCP Access
The new MCP Access experiment makes it easier to connect WordPress to external AI agents through the Model Context Protocol (MCP).
MCP provides a standardized way for AI applications and agents to discover and interact with tools and capabilities exposed by other systems. When enabled, the experiment installs and activates the MCP Adapter plugin when site permissions allow. It also adds a new Tools → MCP Access administration screen where site owners can review registered WordPress Abilities and control which ones are exposed to AI agents.
The interface displays the site’s MCP endpoint and provides controls for exposing or hiding individual Abilities, along with the option to restore their default settings. This separates the underlying MCP infrastructure from the controls site owners need to manage agent access, making it easier to explore how external AI systems can interact with WordPress functionality.
New Experiment: WebMCP
AI 1.5.0 also introduces a WebMCP experiment, allowing compatible AI-powered browsers to interact directly with the WordPress Block Editor.
While MCP Access exposes server-side WordPress Abilities to external agents, WebMCP focuses on interactions within the browser itself. Using the experimental document.modelContext interface, an AI agentAI AgentAn AI system that autonomously performs tasks and interacts with systems through available tools and resources. can interact with the editor through a collection of tools that support actions such as:
Reading the current post and updating its title.
Inserting, editing, moving, and duplicating blocks.
Modifying block attributes and content.
Saving and publishing posts.
These actions use the editor’s existing data stores, so changes appear directly in the editing interface as they happen. This provides an early look at how AI agents could work alongside editors within familiar WordPress interfaces rather than operating exclusively through external APIs.
New Experiment: Revision Change Summary
The new Revision Change Summary experiment helps editors understand what changed between two revisionsRevisionsThe WordPress revisions system stores a record of each saved draft or published update. The revision system allows you to see what changes were made in each revision by dragging a slider (or using the Next/Previous buttons). The display indicates what has changed in each revision. of a post.
WordPress already provides detailed revision comparisons, but interpreting those changes can become time-consuming for longer or more complex content. With this experiment enabled, editors can generate a plain-language summary of the differences between revisions directly from the visual or classic revision comparison screens.
The summary highlights meaningful changes to post content and titles, helping editors, reviewers, and content teams understand how a post has evolved without reviewing every individual change. The experiment is read-only and does not modify either revision or the underlying post. Additional improvements also preserve more detailed comparisons for multiline content such as tables and quotes.
As posts and pages grow in complexity, List View can become difficult to navigate when multiple Group, Columns, and other container blocks appear with generic names. This experiment uses AI to suggest concise, descriptive labels based on the content and purpose of a container. Generated labels are available in List View and the block Rename modal, helping editors identify sections without changing their visible content.
A follow-up enhancement also incorporates document-level context, including the post title, surrounding sections, and existing labels, to produce more useful suggestions.
For editors working with complex page layouts, reusable patterns, or long-form content, these labels can make document navigation and organization considerably easier.
Expanded embeddings infrastructure
AI 1.5.0 continues building on the Embeddings infrastructure introduced in previous releases.
Using WordPress Cron, the system can process existing content in batches, detect when content changes, and update or remove stored embeddings as needed. It also supports resuming interrupted operations and avoids unnecessary requests when content has not changed.
Embedding generation is not automatically activated for every site. Features that need embeddings must register their requirements, allowing the infrastructure to run only when needed.
Like the earlier embedding storage and vector ranking additions, this is primarily developer infrastructure rather than a new end-user experiment. It provides additional building blocks for future semantic search, content recommendations, retrieval, and other AI-powered workflows.
Previously, translation focused primarily on paragraphs and headings. AI 1.5.0 expands that behavior to blocks containing rich-text content, allowing more of a post’s content to be translated within the existing workflow. The release also improves translation behavior when working with the editor’s “Show template” mode and prevents translations from being applied to blocks that have been edited, removed, or replaced while a request was processing.
These changes make Content Translation more useful for complex posts and help protect editors’ work when content changes during a translation request.
Individual posts requested with ?output_format=markdown now include their approved comments, providing AI agents and other Markdown consumers with additional context from the discussion surrounding published content. The release also improves site-wide comment feeds, fixes content negotiation so requests respect the preferred response format, and applies existing WordPress feed filters for greater compatibility with other plugins.
One behavior has changed: individual post URLs using /<post>/feed/markdown/ are no longer supported. Consumers should request the post’s Markdown document using ?output_format=markdown instead.
Improved AI settings and provider discovery
AI 1.5.0 makes it easier to understand how different AI capabilities operate before enabling them.
The AI settings interface now more clearly distinguishes between stable Features and experimental functionality. Experiments that run automatically are identified, and an information control provides details about when each experiment runs, what information is sent to the configured AI provider, and how many requests it makes.
These improvements are intended to give site owners better visibility into AI usage, especially for experiments that can trigger requests automatically.
The release also expands the available connectors featured on Settings → Connectors, adding Mistral and Vercel AI Gateway alongside the existing Anthropic, Google, and OpenAI options. Featured connectors can be discovered, installed, activated, and configured from the same interface.
Additional links to the AI-Powered WordPress course make it easier for new users to find educational resources.
Updated Abilities Explorer
The Abilities Explorer has been rebuilt using WordPress DataViews, bringing its interface more closely into alignment with modern WordPress administration experiences. The updated Explorer makes it easier for developers and administrators to discover, inspect, and execute registered WordPress Abilities.
Three new REST APIREST APIThe REST API is an acronym for the RESTful Application Program Interface (API) that uses HTTP requests to GET, PUT, POST and DELETE data. It is how the front end of an application (think “phone app” or “website”) can communicate with the data store (think “database” or “file system”)
https://developer.wordpress.org/rest-api/ endpoints support retrieving and running Abilities, replacing parts of the previous implementation that relied on older administration patterns. Developers can also extend the Explorer with additional read-only columns using the new ai.abilitiesExplorer.fieldsJavaScriptJavaScriptJavaScript or JS is an object-oriented computer programming language commonly used to create interactive effects within web browsers. WordPress makes extensive use of JS for a better user experience. While PHP is executed on the server, JS executes within a user’s browser.
https://www.javascript.comfilterFilterFilters are one of the two types of Hooks https://codex.wordpress.org/Plugin_API/Hooks. They provide a way for functions to modify data of other functions. They are the counterpart to Actions. Unlike Actions, filters are meant to work in an isolated manner, and should never have side effects such as affecting global variables and output..
Updated core Abilities and breaking changes
AI 1.5.0 includes important updates to three general-purpose WordPress Abilities:
These Abilities have been synchronized with their corresponding implementations coming in WordPress core with 7.2 this December, helping developers work with consistent interfaces as the Abilities APIAbilities APIA core WordPress API (introduced in 6.9) that creates a central registry of capabilities, making WordPress functions discoverable and accessible to AI agents, automation tools, and developers. Transforms WordPress from isolated functions into a unified system. evolves. The updates introduce changes to request and response structures, including new where, fields, and pagination inputs for content and user queries, consistent result structures, and updated field names.
These updates include breaking changes for existing integrations. The deprecated aliases introduced in earlier releases have now been removed:
core/read-content → use core/content-query
core/read-users → use core/users-query
core/read-settings → use core/settings-get
Developers using these Abilities should review the updated schemas and adjust existing calls accordingly. Additional changes to date fields and user lookup behavior may also require updates.
Additional fixes and improvements
The release includes several fixes and smaller enhancements across existing experiments and administration interfaces:
Title Generation: A new Generate title button is available in the post Rename modal.
AI Request Logging: Successful requests from providers using custom transports are now recorded, and administrators can delete logs by age rather than only purging everything.
Suggest Reply: Reply suggestions are applied to the correct comment, and protected post titles and excerpts are no longer sent to AI providers.
Comment Moderation: Automatic analysis no longer processes pingbacks, trackbacks, or block editor Notes by default.
Content Classification: Improved handling of taxonomy terms containing special characters or non-ASCII text.
Key Encryption: Integration with the emerging WordPress Secrets APIAPIAn API or Application Programming Interface is a software intermediary that allows programs to interact with each other and share data in limited, clearly defined ways. includes backward compatibility and migrationMigrationMoving the code, database and media files for a website site from one server to another. Most typically done when changing hosting companies. for existing encrypted keys.
Performance: Feature and experiment settings are loaded more efficiently, reducing repeated database queries on sites without persistent object caching.
AI Status: The dashboard widgetWidgetA WordPress Widget is a small block that performs a specific function. You can add these widgets in sidebars also known as widget-ready areas on your web page. WordPress widgets were originally created to provide a simple and easy-to-use way of giving design and structure control of the WordPress theme to the user. now summarizes enabled and disabled experiments in a more compact format.
The release also improves how private and password-protected post information is handled in AI-powered comment workflows and includes additional protections around stored connector credentials.
Sites using the AI plugin are encouraged to update promptly.
What’s next?
Work is already underway on additional Features, Experiments, Developer Tools, and Abilities planned for 1.6.0 and future releases, including:
As always, we welcome feedback, testing, and contributions from the community. Whether you are interested in editor UXUXUX is an acronym for User Experience - the way the user uses the UI. Think ‘what they are doing’ and less about how they do it., APIs, accessibilityAccessibilityAccessibility (commonly shortened to a11y) refers to the design of products, devices, services, or environments for people with disabilities. The concept of accessible design ensures both “direct access” (i.e. unassisted) and “indirect access” meaning compatibility with a person’s assistive technology (for example, computer screen readers). (https://en.wikipedia.org/wiki/Accessibility), performance, or AI ethics and policy, there are many ways to participate.
You must be logged in to post a comment.