Batesian mimics - palatable mimics of a poisonous species - are detrimental to the species they imitate. Predators learn which species are unpalatable by tasting some. If Batesian mimics are rare or nonexistent, few if any of these samples will be tasty and predators'll quickly learn to leave this species alone. But if Batesian mimics are common, predators have good chance of being rewarded for attacking things that resemble the model species. It would seem that an overabundance of mimics would create a condition for the model species to evolve to visually differenciate itself: is there any evidence that this occurs?
I have found some papers that support the hypothesis that an overabundance of Batesian mimics creates a condition for model species to evolve differentiating characteristics -- in particular, Franks, D. W., G. D. Ruxton, et al. (2009). "Warning signals evolve to disengage Batesian mimics." Evolution 63(1): 256-267.
The authors argue that new conspicuous traits in unpalatable species evolve in response to mimicry, since a mimic that evolves conspicuousness without being a good-enough match pays full cost of conspicuousness without the benefit of its predator learning advantage.
There is an ongoing discussion in the literature on whether Batesian mimicry is always parasitic or can also be mutualistic. I don't know much about this discussion and won't be able to make a synthesis. I suspect semantics about what Batesian mimicry really will also matter in this debate.
In Parus major (and from the same first author), I could find an evidence supporting that Batesian mimicry is parasitic (Rowland et al. 2010) and another evidence suggesting it is mutualistic (Rowland et al. 2007) depending upon the details of the study system. Franks et al. 2009 suggests that in extreme cases, Batesian mimicry can indeed be parasitic.