The free fields are a repackaging of an infinity of de-coupled (normal mode) oscillator operators $a_{\vec p}$,
$$
\phi(\vec x,t)= \int\!\!\frac{d^3p}{(2\pi)^3 \sqrt{2\sqrt{m^2+|\vec p|^2}}} ~
e^{it \sqrt{m^2+|\vec p|^2} -\vec p\cdot \vec x } ~a^\dagger _{\vec p} +\mathrm {h.c.}
$$
in a Lorentz-invariant manner (which you are not asking about), with coefficients specifically chosen in the packaging so these fields obey the K-G equation, manifestly.
The states of the theory consist of infinite products of powers of such oscillators acting on the vacuum, $|0\rangle$, of any and all types (momenta, $ \vec p$): the Fock space. In general, they have no clue about the K-G equation, as they lack the variables it controls, anymore than oil paints have any clue about the Mona Lisa. The K-G equation relates and circumscribes the field operators.
Nevertheless, if you construct states by operating on the vacuum with such quantum fields which depend on t and $\vec x$s, those states will naturally reflect the above arrangement, Lorentz invariance, symmetries, and K-G time-evolution outlined above, in general in a complicated manner of your design. For example, the state $\phi |0\rangle$ will trivially satisfy the K-G equation. Why do you care?