WordPress.com announced today that AI agents like Claude and ChatGPT can now create, edit, and publish blog posts directly on the platform through the Model Context Protocol (MCP).
The update, announced on the [WordPress.com blog](https://wordpress.com/blog/) this morning, represents a significant shift in how content management systems interact with AI tools. Any posts written by AI agents will start as drafts, giving site owners final approval before publication.
What Changed
Previously, using AI for WordPress content meant copying and pasting from chatbots or using third-party plugins with varying reliability. Now, AI agents can connect directly to WordPress.com sites and handle the entire workflow—from drafting to formatting to scheduling.
The integration uses MCP, an open standard that lets AI assistants interact with external systems safely. Think of it as an API specifically designed for AI agents rather than traditional applications.
How It Works
Site owners can authorize specific AI agents to access their WordPress.com site. Once connected, the AI can:
Create new posts and pages
Edit existing content
Manage drafts and revisions
Update meta descriptions and SEO fields
Schedule publications
All changes remain in draft status until a human reviews and approves them. WordPress.com says this safety mechanism prevents rogue AI from publishing unvetted content.
The Claude Connector
WordPress.com previously released a Claude Connector plugin in early February, which laid the groundwork for this broader MCP integration. That plugin has already been used by thousands of sites, according to company data.
Anthropic's Claude was the first AI assistant to gain this level of integration, but the new MCP support opens the door to ChatGPT, Gemini, and any other AI tool that implements the protocol.
Why This Matters
For solo bloggers and small publishers, this could mean more consistent posting schedules without hiring writers. An AI agent could monitor topics, draft articles, and queue them up for review—all while the owner focuses on strategy.
For larger publishers, the automation potential is even bigger. Imagine an AI agent that watches for breaking news in your niche, drafts quick summaries, and places them in your editorial queue within minutes.
But there's a catch: quality control becomes critical. AI-generated content still needs human oversight to catch errors, maintain brand voice, and verify facts. WordPress.com's draft-by-default approach helps, but editors will need clear workflows for reviewing AI submissions.
The Automation Debate
Not everyone's thrilled. Some WordPress community members worry this will flood the internet with more low-quality AI content—what some call "slop." Others see it as a natural evolution of content tools, no different from spell checkers or grammar assistants.
WordPress.com hasn't released usage statistics yet, but early adopters report saving hours per week on routine content tasks. The trade-off is investing that saved time in editing and quality assurance.
Getting Started
WordPress.com users can enable MCP integration through their site settings. The feature is available on all plans, though rate limits may apply to free accounts.
Documentation for developers building MCP integrations is available on the WordPress.com developer portal.
Sources:
[WordPress.com official blog announcement](https://wordpress.com/blog/) (March 20, 2026)
[The Verge AI coverage](https://www.theverge.com/ai-artificial-intelligence) (March 20, 2026)
