There was an update of the XMP library (Adobe XMPLib) used by Automation Engine (Classic) tasks and Imaging Engine. This library is used to write XMP metadata in (Normalized) PDF files, TIFF/LEN,... files. With this update, the structure of the generated XMP has changed. Previously, different sections within the XMP all had their own rdf:Description element. Now all sections are combined in one rdf:Description element.
All files written by an Automation (or Imaging) Engine version 22.11 or higher will have simplified but different XMP metadata.
This can have an impact on SmartNames that queried the XMP metadata (XMP XPath Query SmartNames). SmartNames that relied on the old structure will have to be adapted.
In the old structure, layer information had it's own rdf:Description element:
To list all layer names, one could have created an XPath like:
The predicate  for the Description element tells the server to look in the 9th Description element specifically. As everything is now gathered in one Description element, there is no 9th element and the query will have no result.
Although the use of such a predicate was not a good idea in this case (not all files will have the layers section in the 9th element), this might have worked for specific cases.
The solution here (and typically a better approach in general) is to remove the  predicate: