Is it safe to study from MIT and Berkeley course series, or they contain wrong information? After surveying most of the universities introductory physics courses, I found none is using Berkeley physics books or MIT physics books as textbooks. All are using Halliday, or Serway and the like.
What is the reason behind not using these books anymore (once they were the standard textbooks, may be before halliday I suppose) ? Do they contain old/outdated/wrong information or something?
Did Berkeley and MIT develop a new set of books to substitute those books? 
 A: It's very slightly incorrect to say that "nobody" uses them. My own undergraduate university used Purcell for the more mathematically advanced section of the first-year course. We also used Reif for Thermodynamics and Stat Mech (NB: my copy of Reif suggests that it's not part of the Berkeley series, and I think I remember another book Reif wrote, but I don't remember it). 
There certainly isn't any incorrect content that I'm aware of in either of those series. The material is all old enough and well-established enough that there isn't going to be any revolutionary changes necessary to the books. Purcell's book is more math-heavy than Halliday and similar books, so that's probably the reason it's not used for intro classes. As I said earlier, we used it for the more mathematical version of the intro sequence. There was also a version of the sequence based on Halliday. The version I took is arguably not "introductory," which is why I say "slightly" incorrect above. We also did Thermo/Stat Mech in fourth year. Reif's book could probably be taught to more introductory students, but it's still arguably not introductory (it doesn't really rely on much other physics, but it does use partial derivatives, for example). 
I'm not familiar with the rest of the Berkeley series. My only familiarity with the MIT series is that I once skimmed a copy of the Special Relativity book in that series. My best guess for the reason that the are rarely used is that they are mathematically harder than departments want to use, and/or that few departments want to use a separate book for special relativity.
EDIT: Both the Berkeley series and the MIT series appear to date from the 1960's, and are actual full-length textbooks. Both departments commissioned their series to teach the material the way they thought it should be taught. For the Berkeley series, all the authors are from the Berkeley Physics Department. MIT had Anthony French write all their books, and Edwin Taylor co-authored the quantum book. Both of them are at MIT. The name of the series does not mean that that department actually uses that series; it just means that they caused it to exist. Berkeley appears not to use any of their books at the moment; that information is harder to find for MIT. 
@Ron Maimon's mentions the Feynman lectures and Landau and Lifshitz in his answer; the Berkeley and MIT series are the same idea. It would probably not introduce any major inaccuracies to think of the Feynman lectures as the "Caltech Series" and LL as the "Russian Series."
A: It is a mistake to assume that the books that are standardized are the ones that are superior, or have fewer mistakes. Generally, they have fewer typos, because they get more exposure, but the rate of typos and errors in most undergraduate books is roughly constant across the board.
If you find that a school is using a different set of books then the standard ones, and these are not written by local authors, so it isn't nepotism, this means somebody at the school put some thought into designing the curriculum. This is usually as sign that the books are superior.
So you should use the MIT physics books, as they are probably pretty good. Although to be honest, I think that nothing in the American physics system, with the possible unique exception of the Feynman's lectures, can compare to the Soviet era Russian books, in particular the Landau and Lifschitz series, which is superb in both scope and quality.
Purcell is a decent book, although it uses CGS, which takes a bit of getting used to. I don't know the other books. It is better to just ignore the majority opinion when considering the quality of books, since majority opinion tends to bury classics and elevate mediocrity. Popular opinion decides how true something is based on how familiar it sounds, and this is not a procedure which rewards originality or vision.
