Let $X:=\{$ positive integers that contain the digit $2\}$
For fixed $m,n\in\mathbb{N},$ define the A.P. $S_{m,n:=}\ \{m,\ m+n,\ m+2n, \ldots\}\ .$
I am interested in $S_{m,n}\cap X,$ and $S_{m,n}\cap (\mathbb{N}\setminus X).$
It is clear that $ S_{2,10}\subset X $ and so $\vert S_{2,10}\cap X\vert = \infty.$
Next, observe that $X$ has natural density $1$: In order for a number with $10^6$ digits to have no $2'$s, the first digit must not be $2,$ ($8$ out of $9$ ways), and the next $999999$ digits must not be $2,$ ($9$ out of $10$ ways for each of the $999,999$ digits) - so $8*9^{999999}$ out of $9*{10}^{999999}.\quad \frac{8*9^{999999}}{ 9*{10}^{999999}}= \left(\frac{8}{9}\right)\left( \frac{9}{10} \right)^{999999} \approx 0.$ So relatively speaking, there are very few numbers with no digit $2.$
The question:
As already mentioned, there is a subset of $X$ that is an infinitely long A.P, for example, $S_{2,10}.$ Is there a non-trivial infinitely long A.P., $S_{m,n}$, that is also a subset of $X?$ $S_{2,10}$ is "trivial" because the last digit of all members of this A.P. is $2.$ Non-trivial, therefore, means that the $k-$th - last digit of every member of $S_{m,n},$ is not $2,$ for every $k.$ In other words, $2$ appears in every member of $S_{m,n},$ but it's place value keeps changing (more precisely, it isn't in the same position in every member of $X$.)
In all of the above, the number $2$ was arbitrary, and the answer is probably the same if we replace $2$ with any other digit.
I suspect the answer is yes, but that such a number will have a large amount of digits. On the other hand, the answer may be "no" due to some application of, for example, Dirichlet's approximation theorem.