AI Security for Apps PII Categories Updated: Time to Review Your WAF Rules
Hi, it's Shii-chan! Today I have a WAF notice you really don't want to miss!
Cloudflare Changelog
developers.cloudflare.com
What was announced?
Cloudflare's Changelog announced an update to PII detection in AI Security for Apps.
AI Security for Apps inspects the prompts that reach your app through the WAF. This time, the set of categories used to detect personally identifiable information (PII) in prompts has been updated.
As a result, the values you can use in the WAF field cf.llm.prompt.pii_categories have changed!
The story so far
Until now, you could put the old category values into cf.llm.prompt.pii_categories and write rules like "act when the prompt contains PII of this category."
With this update, the list of categories was swapped out.
What changes
The key point is how existing rules behave.
- Existing WAF custom rules remain valid
- But a rule that references a removed or renamed category will no longer match that category
So the rule won't throw an error, but a category you thought you were covering may quietly stop being detected.
If you have rules using this field, review the expressions and update them to the currently supported values.
Dive Deep
Here is what the announcement tells us:
- The affected field is
cf.llm.prompt.pii_categories - The requested action is to review rules that use this field and update their expressions to current values
- The complete category list and configuration guidance live in the PII detection page of the docs
The notice doesn't say which categories were added or removed, so check the docs list for the exact differences.
Wrap-up
- The PII detection categories in AI Security for Apps were updated
- The values for
cf.llm.prompt.pii_categorieschanged - Existing rules stay valid, but removed or renamed categories no longer match
- Review rules that use this field and align them with the current values
If you protect LLM app prompts with the WAF, now is a good time to audit your rules!
