WordPress Now Shows You Your AI Bill. Here's How to Actually Control It.
WordPress 7.0 ships with an AI Request Logs dashboard — a React-powered interface under Settings → AI Request Logs that shows you every outbound AI call your site makes: the provider, the model, the token count, how long it took, and an estimated cost.
That’s real progress. Knowing what your site is spending on AI is the first step to managing it. Until now, that data wasn’t in WordPress at all — you had to dig through your provider’s billing dashboard and try to reverse-engineer which plugins were responsible for which charges.
But visibility and control are two different things. And the gap between them is where most of the actual risk lives.
What the AI Request Logs Dashboard Does
The feature shipped as part of the WordPress/ai plugin, developed across PRs #149 and #437. The PR description is clear about what it’s designed for: “observability surface for debugging, cost tracking, usage analysis.”
That framing is deliberate and accurate. The dashboard is a read-only view. You’ll be able to see:
- Which AI provider handled each request
- Which model was used
- Token counts per request
- Request duration
- Estimated cost per call
This is genuinely useful. If your AI bill spikes unexpectedly, you’ll finally have a place inside WordPress to start diagnosing it. Developers troubleshooting why a plugin is slower or more expensive than expected will find this invaluable. Site owners doing a quarterly cost review will have actual numbers to work with.
What the dashboard doesn’t do — by its own design — is stop anything from happening.
The Gap: Observation Is Not Enforcement
Here’s the scenario worth thinking through.
You open Settings → AI Request Logs on a Tuesday morning and see that your WooCommerce product description plugin ran 12,000 tokens on Monday. That’s more than you expected, but it’s within range. You make a mental note and close the tab.
Wednesday it runs 45,000 tokens. Thursday it runs 200,000. By the time you’re back at your dashboard on Friday, the damage is done.
The logs told you everything that happened. They couldn’t stop any of it.
This is the structural limitation of an observability-only tool when the underlying problem is an unbounded process. Knowing you got a big bill doesn’t prevent the next one. What prevents it is enforcement: a rule that says “this plugin gets 50,000 tokens per month, and when that’s gone, it stops.”
The WordPress/ai plugin’s logging feature isn’t trying to be that. It’s explicitly scoped to observability. That’s a reasonable design decision — you have to build visibility before you can build controls on top of it. But it means the work of actually controlling spend remains unsolved.
What Enforcement Actually Requires
A complete AI spend control system for WordPress needs three things that go beyond what a request log provides:
Per-plugin token budgets. Each plugin should have its own monthly allocation. When that allocation is exhausted, that plugin’s AI calls stop — without affecting anything else on the site. A bulk job in one plugin shouldn’t drain the budget for every other AI feature running on the same site.
Hard stops, not just visibility. An audit log shows you what happened. Hard stops prevent what’s about to happen. These are different mechanisms. The wp_ai_client_prevent_prompt filter — a hook that fires before every outbound AI call — is the right interception point for enforcement. It already exists. What’s missing is a tool that uses it to apply budget rules automatically.
Per-plugin audit accountability. The new logging dashboard shows you requests. But for enforcement to work, those requests need to be attributed back to specific plugins so budget rules can be applied per-source. “Your site made 400 AI calls yesterday” is useful for debugging. “Plugin A made 380 of them and is over its monthly budget” is what you need to act.
The logging dashboard lays the groundwork. It’s collecting the data that a budget enforcement layer needs. But that layer has to be built separately.
How the Two Layers Fit Together
It’s worth being clear that Core’s logging feature and a governance layer aren’t competing — they’re complementary.
The WordPress/ai team is building the observation layer: a clean, React-powered view of what your site’s AI calls look like in real time. This is the right thing for Core to ship. It’s useful for every kind of user, from developers debugging a plugin to store owners trying to understand an unusual bill.
A governance layer — the enforcement half of the equation — does something different. It uses the same request data, applies per-plugin budget rules, and blocks calls that would exceed those rules. It operates at the wp_ai_client_prevent_prompt hook before the request leaves your server, not after it’s already been logged.
The two layers together give you the complete picture: see what’s happening (Core), stop what shouldn’t be (governance). Neither one is the full solution on its own.
Where We Come In
We built that enforcement layer.
The AI Governance Plugin for WordPress sits on top of the native AI client and adds the controls that the logging dashboard deliberately doesn’t include: per-plugin token budgets, hard stops when those limits are hit, and attribution that maps each request back to the plugin that triggered it.
The architecture is built around the hook WordPress already ships. The wp_ai_client_prevent_prompt filter fires before every outbound AI call. AI Governance Pro uses it to apply your budget rules before anything gets sent — not after it shows up in a log.
AI Governance Pro is available now at $39/year. For most WooCommerce stores, that’s less than the cost of a single runaway overnight bulk job. See pricing at axtolab.com/pricing →
The Takeaway
WordPress 7.0’s AI Request Logs dashboard is a real step forward. Seeing your AI spend broken down by provider, model, and token count — from inside your WordPress admin — is something the platform has needed since the shared AI client shipped.
But observation is not control. Knowing what you spent last week doesn’t prevent what you spend next week. For sites with meaningful AI workloads, production WooCommerce stores, or any situation where a rogue bulk job could generate an unexpected bill overnight, visibility alone isn’t enough.
The logging layer tells you what happened. AI Governance Pro stops what’s about to happen.
AI Governance Pro enforces what WordPress 7.0 only shows you. $39/year →