Well, the simple answer is "YES", but whether you believe me or not is up to you.
When I was the Reference Service Manager of the Red Cell Immunohaematology (RCI) Laboratory at the Tooting Centre of the National Health Service Blood and Transplant (NHSBT), we had a patient's sample referred to us from one of our samples from the East Coast of England (I have to be careful not to identify either the patient or the hospital) who had an anti-K, having never been transfused with K+ blood. However, this patient consistently had a positive anti-K in their plasma, and also, believe it or not, could have anti-K eluted from their erythrocytes,
Knowing the situation (i.e. we had not supplied K Positive blood to the hospital for this patient for many years, AND knowing that they knew what they were doing - they would NOT have given K Positive blood), I was wondering if either I, and/or my staff (in their case, almost impossible, even if I was fallible) and so we sent the sample to the International Blood Group Reference Laboratory (IBGRL) for confirmation. The report we got back (from Joyce Poole) was that they also detected an anti-K, from an apparently K Negative patient with a positive DAT, but the eluate was (again, apparently) anti-K!
Unfortunately, we lost track of this patient, BUT, if Joyce was a bit foxed by this case, I feel TOTALLY free to be foxed as well! Her theory was that this was a case of a "mimic-anti-K", rather in the same way of almost all WAIHA specificities being a mimicking "specific Rh antibody". Since then, of course, it has been shown that low prevalence antigens within the Kell Blood Group System can lead to "strange" antibody specificities within the Kell Blood Group System, together with weakly expressed antigens within the Kell Blood Group System.
I am NOT saying this is a total answer to your query (but it is the best I CAN DO!).