Philosoraptor proves once again why it’s vitally important to keep a few dope-smoking philosophers around: The Niger uranium fiction and how the French tried to warn us about it [Oh, and why Bush's "sixteen words" were not even "technically correct"... again].
Anyway, once again: ‘Learned’ is a “success term.” You can’t assert of S that he has learned that p if your best evidence indicates that he’s made a mistake. (In fact, you can’t even assert it if your best evidence is equivocal and forces you to suspend judgment on the matter.) In such cases all you can assert is something like S has come to believe that p.
The situation is analogous to the case of knowledge claims. If my best evidence indicates that p is false I can’t honestly assert that you know that p, but only that you believe that p. If my evidence indicates that p is false (or even that we can’t tell whether or not p is true), but I know that Smith has (so far as I can tell) mistakenly come to believe that p, but I tell you that Smith knows or has learned that p, then I am deceiving you.
And that’s what Bush did.