It doesn't exactly contradict it, because to contradict a model you need to show it does not occur in string theory, and that's going to be a hard slog through vacua. Almost any field theory is known to occur in a configuration of string theory, but as a low-energy limit near a brane stack, not as a theory of the entire universe.
If you try to make a theory of the entire universe using string theory rishons, you will probably get extra stuff that would rule out the theory, like extra generations of rishons, or with too many branes to allow for a high Planck scale geometry, or incorrect gauge fields to make the observed particles, or no gauge coupling unification to allow the forces to come from one place. But then you could try to engineer the system some other way. Using a large-extra dimensions, you will certainly be able to make something qualitatively like this, but you will certainly not match experiment, because the large extra dimension low Planck scale will introduce non-renormalizable corrections which will require insane fine-tuning (as always).
In E8 heterotic compactifications, it is extremely natural to have quarks and lepton generations of the right type produced directly, from a very simple and very beautiful idea due to Witten of putting SU(3)xE6 into E8, and identifying the SU(3) with the holonomy of the Calabi Yau. This idea automatically produces supersymmetric standard-model like low-energy theories, with a ton of generations. If you embed a rishon model, you would need a new idea for getting out the rishon interactions, and you might get several generations of rishons.
But in general, string theory moots the reason people proposed preon models in the first place: explaining generation structure. Standard model generations in string theory are the Euler number of the Calabi Yau, and it is actually reasonably doable to get 4,6,8, or 3 generations without forcing things too much. The fact that heterotic compactifications naturally explain generation structure is one of the strongest bits of qualitative evidence in favor of traditional heterotic-like string compactifications.
Modern ideas can get engineer pretty much any gauge theory and any matter using enough stuff. There were exceptions when I last looked, because you needed branes to stack in too many perpendicular directions for some gauge groups, but close enough for all practical purposes. So I am sure that are brane-world rishon theories, although these are theoretical exercizes having nothing to do with experiment.
I would expect a theory of open strings, with Chan-Paton charges at the ends, to be interpreted very much like a theory of preons. It could be claimed that the points at the end of the string are only "mathematical preons".
Before the eightfold way organized the then (1962) known resonances studied in scattering experiments neutrons and protons were considered elementary alongside electrons. Elementary particle physics was the study of scattering and the study of pole exchanges, Regge poles. One could say that there were no preons.
With the discovery of the symmetries in the data and the emergence of the quark content of the nucleon, one might have said that the quarks were the preons, as far as the compositness of the "elementary nucleon". Preons are another hypothetical level in a nesting sequence. There exist various models of compositness for quarks and leptons which at the moment have no experimental confirmation.
Preons are not a theory but a set of various models, practically numerology, based on a hypothesis of compositness, of how the symmetries observed may arise from a substructure in what we consider now elementary particles.
String theory on the other hand is not a model, but a theory. A model may be embedded in a theory or not, may be compatible or incompatible with a theory but cannot contradict it.
If the Standard Model is embedded in String theory and if ( a big if) a compositness model emerges experimentally that also includes the Standard model where would the contradiction be?