There is a strong temptation to sensationalize science for public consumption. New sources do this constantly (see "Einstein was wrong!" from the faster than light neutrino fiasco last year) but they aren't the only ones. Entertainment is unfortunately full of relatively harmless pretend science from Star Trek to CSI where reality is ignored but nobody really believes that stuff is real and if they did ... so what? Dr. Oz, though, is another thing entirely.
The real trouble is that if people like Oz report the real science behind medical news it sounds really boring. They have to include a lot of 'in animals only, might not apply to humans at all' and 'insufficient sample size' and 'proven to affect markers of disease, but not the actual disease'. All of these things are completely critical to evaluating the efficacy of a treatment but make crappy television so demagogues like Oz just make stuff up instead. Nobody is going to listen to 'needs more research' when 'instant miracle' is available. We don't need miracles. We need boring old broccoli and walking!
Every week Dr. Oz appears on the cover of Woman's World magazine at the grocery store pitching a new and guaranteed instant weight loss program. You would think that after he did this 30 times in a row that people would wonder why they listen to a guy who is so consistently wrong but apparently that doesn't happen. He pitches unproven medicine, supports charlatans, and generally does his level best to completely undermine people's understanding and respect for actual medicine.
The thing that the average person just doesn't get is that science and research is full of boring. It contains lots of doubt and uncertainty. At the end you can have spectacular and amazing but an awful lot of the time all you have is data that shows that nothing new or interesting is happening. That is useful and necessary but it isn't getting anybody's attention, while anti oxidants and fat free diets are miracle quick fixes that are totally useless in reality. Sadly the thing that should really make people believe a statement is true is bunch of complicated and (to the average person) indecipherable references to study size, statistical sampling, and long term side effects while instead it just puts them to sleep.