Whether sulfur or phosphorous actually expand their octet is contested within the chemistry community. Another term for this octet expansion is "hypervalency." You can find many works of research regarding hypervalency. The consensus, according to Wikipedia, is that both can expand their octets, but not to a significant extent. In other words, d-orbital utilization is low at best.
Regarding sulfur trioxide, this article may be of interest.
We find no evidence to support notions of p π –d π back-donation from oxygen to sulfur.
Regarding the octet concept, it's only a guideline at best, and nothing more. I wouldn't put too much stock into the concept. If you insist on using it, then only apply it to the following elements: $\ce{C, N, O}$, and $\ce{F}$. And note that for carbon, there are well known exceptions in organic chemistry - carbocations (which have less than an octet of electrons). Carbocations however tend to exist mainly as reactive intermediates. Therefore, a more nuanced statement might be $\ce{C, N, O}$, and $\ce{F}$, when involved in neutral and thermodynamically stable compounds, generally have 8 valence electrons.
this makes the concept of octet somewhat incorrect.
If anyone can think of exceptions, please share || $C_7H_7^+$ , $C_8H_8^{2-}$ are well-known examples =).
– permeakra Jun 26 '14 at 18:50