Do gravitational lenses act as prisms? Light creates gravity, and the greater the light's frequency, the greater this gravitational effect is. It stands to reason then that light of different colors would react slightly differently to gravitational fields. Namely, bluer light would bend more than red light, causing gravitational lenses to act like prisms.
 A: In general relativity, any test particle's worldline (whether the particle is massive or massless) is determined entirely by its initial four-velocity $u^\mu$, and not by any other properties.  This is manifested by the geodesic equation
$$
\frac{d u^\mu}{d\tau} = \Gamma^{\mu} {}_{\rho \sigma} u^\rho u^\sigma.
$$
From the properties of ordinary differential equations, you can see that if you know $u^\mu$ at some initial time, then it determines the value of $u^\mu$ for the future of the particle.
Intuitively, you can think of this as a consequence of the equivalence principle, extended to photons.  A feather and a hammer released from rest will follow the same trajectories, though they have different masses—which is to say they have different amounts of energy (including rest energy.)  If you accept that photons also follow spacetime curvature in a similar way, then it should seem plausible that the energy content of a photon shouldn't affect its trajectory either, and that two photons with the same energy and the same initial four-velocity will have identical trajectories.
It is true that the test-particle limit is an approximation (both for the geodesic equation, and for the hammer-feather equivalence principle test);  if we take the gravity of the photon/light wave into account, there would probably be some minuscule difference in the scattering angles due to the effect the photon has on the body creating the gravitational field.  But I would bet a sizable amount of money that such effects will not be detectable in my lifetime.  The test particle approximation is an exceedingly good approximation for quite a few gravitational physics situations, including this one.
