This has a precondition: the existence of targets unique to the cancer cell. That's not necessarily true. In theory you can have cancerous cells that are (at least in the epitope-space) indistinguishable from normal cells. I'm not a doctor, and I guess (and hope) this is an usual case, but finding a truly unique and dissimilar mutation sounds pretty hard, training the immune system to detect it correctly without generalizing sounds even harder. Potential, sure, and I hope this proceeds and yields some therapies that are safer than chemotherapy, but we need to be realistic with what we'll probably get.
It's not that hard, in patients with cancer, it's very common for the immune system to go after the cancer specific mutations, until the cancer learns to express the cell-surface checkpoints that deactivate the immune system.
Its extremely rare (or at least I have not heard of it) to develop auto-immune activity during the course of cancer.
T-cell receptors that recogize HLA-bound epitopes tend to be extremely specific, and rarely go after other parts of the body. There's nothing special about cancer in this regard.
