https://www.w3.org/WAI/WCAG22/Understanding/error-identification.html currently lists two techniques which, arguably, aren't really all that "sufficient"

• ARIA2: Identifying a required field with the aria-required property - this, on its own, has nothing to do with error identification. making sure that a field programmatically exposes that it's required doesn't do anything to identify errors. it may help prevent them, but not identify them. at least this isn't just using aria-required, but also visible text as part of the label, so it will work for more than just AT users. the similar PDF5: Indicating required form controls in PDF forms is different as here the required property will also trigger an error validation for non-AT users. maybe this ARIA2 technique should be replaced with one that uses the actual required (rather than aria-required) attribute, which at least will trigger browser-native validation?
• ARIA21: Using aria-invalid to Indicate An Error Field - this technique only works for assistive technology users, in isolation. it won't help any user who isn't using something (generally AT like a screen reader) that also conveys the aria-invalid state, so relying solely on this won't be sufficient for all users