The vulnerability, identified as CVE-2025-13082, is a content spoofing issue in Drupal core. The root cause is the reflection of unsanitized user input from the requested URL format into HTTP exception messages. When a user visits a specially crafted URL with a malicious format specifier (e.g., containing HTML tags), Drupal's exception handling mechanism includes this raw format string in the error page sent back to the user's browser. The browser then renders this injected HTML, allowing an attacker to deface the page or present misleading information.
The security patch addresses this by removing the dynamic, user-controlled format string from the exception messages in three key locations. Instead of reflecting the input, the updated code now returns static, generic error messages. The identified vulnerable functions are the exact locations where this reflection occurred. Any runtime profile during exploitation would show one of these functions throwing either a NotAcceptableHttpException or a BadRequestHttpException just before the malicious content is rendered.