Your security plugin is blocking AI crawlers

What we observed

We read the configuration of a security plugin installed on your site and found rules that name AI crawlers directly. We report the crawler, the reason saved with the rule, and where the setting lives. We do not change it — another plugin’s settings are not ours to touch.

Why it matters

These rules are often inherited rather than chosen. They arrive from a “block bad bots” preset, a blog post, or a support suggestion, and they are rarely revisited. Many site owners blocking GPTBot today have no wish to disappear from ChatGPT — they simply never realized that blocking training and appearing in answers are separate choices.

There is also a second, sharper problem: security plugins usually trust requests coming from the server itself. Our check runs from your server, so a rule can be perfectly active and still not affect our test. That is why this finding reports what the rule says rather than claiming to have proved what it does.

How to fix it

Wordfence: Firewall → Blocking. Existing blocks are listed with their type and reason. A user-agent block appears as Advanced Block with a pattern such as *GPTBot*. Delete the rule, or narrow it so it only matches the crawlers you really want blocked.

Solid Security (formerly iThemes): Security → Settings → Banned Users, and check the “Ban Hosts” and “Agents” lists.

All In One WP Security: WP Security → Firewall → Blacklist Manager, under user agents.

Sucuri: the blocking happens in Sucuri’s own dashboard rather than in WordPress.

Cloudflare: Security → Bots, plus any custom WAF (firewall) rules. These settings live in Cloudflare, so nothing in WordPress can change them.

Deciding what to keep

Before deleting anything, decide which of these you actually want, because they are different crawlers doing different jobs:

Bot	What it does	Blocking it means
GPTBot	Collects pages to train OpenAI models	Your content is not used for training
OAI-SearchBot	Fetches pages so ChatGPT can cite them	You disappear from ChatGPT’s answers
ClaudeBot	Training, Anthropic	Not used for training
Claude-SearchBot	Fetches pages so Claude can cite them	You disappear from Claude’s answers
PerplexityBot	Fetches pages for Perplexity answers	You disappear from Perplexity

Blocking training while allowing answering is a reasonable choice. Blocking answering crawlers is almost always an accident.

How to check it worked

Re-run the check in the plugin. The finding should disappear from the Overview screen. If the plugin records blocked hits for that rule, its counter should also stop increasing.