The idea of taking the converse probability sounds good to me. Let $r,b,w$ the events where no red, no black and no white balls are drawn.
Then it is asked for
$1-P(r\cup w\cup b)=1-\left[P(r)+P(w)+P(b)-P(r,w)-P(r,b)-P(w,b)+P(r,w,b)\right]$
For $P(r\cup w\cup b)$ the inclusion exclusion principle is applied.
$1-\left({5 \choose 0 }\cdot {13 \choose 5}+{6 \choose 0}\cdot {12 \choose 5}+{7 \choose 0}\cdot {11 \choose 5}-{7 \choose 5}\cdot {11 \choose 0}-{6 \choose 5}\cdot {12 \choose 0}-{5 \choose 5}\cdot {11 \choose 0}+0\right)/{18\choose 5}$
$\approx 70.67\%$
The number of binomial coefficients which has to be calculated is more or less the same like in barak manos answer. Both ways lead to the same result.