If they do it right, this is actually a wise approach: test it. Does the keto diet reduce problems of diabetes and other blood sugar issues? If so, does it do that better/faster/cheaper/safer than other methods? Does it cause other problems? If so, are those less bad, the same, or worse than problems caused by diabetes and/or other treatments such as drugs? You have to look at the net effect, not just changes in one area like blood sugar itself. I don't know whether the VA is being that careful. I do know that testing things, preferably with a large number of diverse willing participants, is much better than just taking it for granted that something is "good" or "bad" or "effective."
As a general rule, eating lots of carbs and fat tends to cause problems with blood sugar. Keto is low-carb but high-fat. Will it perform better or worse than other diets, for example, those with lower fat and carbs that rely extensively on fresh produce? My suspicion is keto will perform worse, but some people might find it works better for them. I'm curious about the results. I keep an eye out for studies like this, because people have a tendency to run after this or that fad diet instead of actually measuring what it does. Science matters.