CRISPR-cas9 uses a string of RNA that matches with DNA and makes a double stranded cut at that point.
If the RNA is just a few letters in length, the enzyme would cut DNA in many places. It would be unspecific. But if you can make the RNA string very long, it would only cut at the exact place where you want to cut.
So how long can you make the RNA sequences? And is this enough to avoid unintended cuts?
The problem of off-targets in CRISPR/Cas is often discussed. It was shown that the system allows mismatches up to five basepairs. For your question, if it is helpful to elongate the gRNA: it was shown that truncating the RNA enhances the specificity more than elongating (see also here).
So what can we do? Well, there are different methods to improve the specificity. The simplest may be the usage of lower dosages of the CRISPR/Cas system. Another method is to use altered PAM motifs.
A more complicated method is to convert the Cas9 into a nickase. In this strategy you would use two different gRNAs for two different target sites and introduce ssbreaks. A modification of this method is to completely knock-out the enzymatic activity of Cas9 and fuse it to a FokI nuclease which will cut the DNA only when it dimerizes.
There are several advantages in these methods, however, you may also loose on-target effects.
There is another method which, in my opinion, is really awesome. It is possible to alter the energetics of the DNA binding site resulting in high-fidelity variants of Cas9 (HF1 and HF2)
So, I hope I could help you. There are other and even more effective ways to increase CRISPR/Cas specificity instead of altering the gRNA. If you're interested, have a look on the references.
