People who browse the web for medical information often cannot discriminate between scientifically valid information ...

Why does the negation test not work for A (People who browse the web for medical information do NOT do so to diagnose medical conditions)? If the conclusion is that they will do more harm than good, then won't this make the argument fall apart?