Don't read too much into "Light can be slowed down or stopped". Be careful of such statements - they are either journalistic "sensationalism" or the valid effort of science writers trying to explain the properties of exotic optical materials in everyday words.
Firstly, in general relativity, measured lightspeed can vary (see, for example, Wikipedia's Rindler Co-ordinates article) in a frame with proper acceleration, so there the variation is to do with the observer's whim (he or she decides to go crusing in a rocket and blasts off relative to an inertial frame. Or, more mundanely, they decide to sit on the surface of a planet and insist on doing GR calculations in the Rindler metric even though their accelerometer screams at them that there is a more "natural" frame to do their analysis in). OK, I'm being slightly flippant - the Rindler metric has a valid use simplifing some analysis but the seemingly nonconstant lightspeed (nonconstant co-efficient $g_{00} = g^2 x^2$ in the fundamental form) it introduces is a property of the deliberately chosen co-ordinates, not of the physics - in the same way that the superficial "bendiness" of spherical or cylindrical polar co-ordinates labelling Euclidean $\mathbb{R}^3$ belongs wholly to those special co-ordinate systems, and not to the underlying flat Euclidean $\mathbb{R}^3$ space.
Secondly, there is a great deal of interest in "slowing light down" for the purposes of optical storage and such like. All optical mediums made of matter (i.e. as opposed to the vacuum) "slow light down". Call me pedantic, but this really isn't "light" in the fundamental sense. I like to think of such things as quantum superpositions of free photons and excited matter states: an optical medium absorbs photons and then re-emits another "elastically" (i.e. in the same momentum, energy and angular momentum state) a short time later (this happens through interaction with the material's electrons) thus seemingly, through this delay, slowing the light down. "Slow light" research is simply finding really exotic versions of this mechanism so that pulses can be held in storage for either exotic switching purposes in a telecoms network (optically re-ordering a data packet) or keeping them idle if needed for data processing or quantum computing.
But these valid ideas of "slow light" don't affect $c$ as we know it. As for whether this $c$, as the fundamental speed of a massless particle for all observers, has been different in the past or whether it might be so in the future: this is the meaningful part of your question, whose answer I shall defer to a cosmologist. To my untrained eye, if General Relativity is right, the Universe and its history is supposedly a manifold solving the Einstein Field Equations (there is ONE manifold with 3+1 dimensions, not an evolving 3-manifold with a history), and my understanding of the physical content of GR is that it roughly says that tangent spaces to this manifold constructed in a certain way from geodesics are inertial frames wherein special relativity holds, so, by definition, wherever and whenever general relativity holds, $c$ has to be constant. So we'll need a real general relativity-ist to confirm this understanding and cosmologist to answer where, when and how GR is seriously taken to falter.
