This is not an easy question to address. I will outline what Hawking radiation is and how it works. I can then point to what the plausible difficulty with it lies. It is a semiclassical theory in that it treats the black hole as a classical system that emits quanta of radiation. The adjustment of the black hole to a smaller mass by a tiny increment is treated with a metric back reaction.
If you have a particle on an accelerated frame it is within what is called a Rindler wedge. Below is a diagram of spacetime for an accelerated frame.
We think of a particle in region I. The hyperbolic lines are regions of constant radius from the $45$ degree lines that are a particle horizon. An observer on an accelerated frame, with acceleration $g$, has the observer behind them at a distance $\rho~=~c^2/g$. The larger the acceleration the the closer to the horizon the observer is. It is also interesting that for two particles to remain a constant distance from each other they must have different accelerations.
One observer, call him Bob, in region $I$ is not able to ever observe anything in region $II$, say if there is an observer named Alice there, or communicate to region $II$, and Alice in region $II$ is not able to communicate to Bob.
The spacetime metric distances are parameterized as
$$
t~=~\rho sinh\omega,~x~=~\rho cosh\omega
$$
the angle $\omega$ is a parametrized time. the metric in the Minkowksi form is then
$$
ds^2~=~-d\rho^2~~-~\rho^2 d\omega^2~-~dy^2~+~dz^2.
$$
If we euclideanize this so that the metric is not Lorentzian we can then think of the unitary time development operator $U(t)~=~exp(-iHt)$ across region $I$ to region $II$. We do this to consider the evolution of a quantum fluctuation that encloses the origin of the diagram above. We then replace $i~\rightarrow~1$ and the time is evaluated for the entire loop, think of this as the perimeter of the loop, as $t~\rightarrow~\rho\omega|_0^{2\pi}$ $=~2\pi\rho$. We then have the operator $U(\omega)~=~exp(-2\pi\rho H)$.
Alice and Bob measure the quantum fluctuation, say a loop that encloses the origin, as a particle that emerges from the horizon and then approaches it again. The particle emerges from the past horizon slowly and then slowly approach the future horizon, for Bob in region $I$ can only observe in a redshifted and time dilated form. Alice in region $II$ observes the same. For this virtual loop we may think of Bob and Alice as witnessing different states $\phi(b,b')$ and $\chi(a,a')$, but which form an entangled state $\psi$ with density matrix $\rho_{AB}~=~\psi^*\psi$
$$
\rho(a,a',b,b')~=~\chi^*(a,a')\phi^*(b,b')\phi(b,b')\chi(a,a'),
$$
where Alice and Bob observe what can be found by tracing over Bob's and Alice's state variables $b,b'$ and $a,a'$.
The time evolution operator has become a thermal or Boltzmann operator. The temperature is then $\beta~=~2\pi\rho$ or
$$
T~=~\frac{1}{2\pi\rho k_B}.
$$
This is the Unruh effect, explained in elementary terms. The Rindler wedge has no curvature. A black hole of course has curvature; the Riemann curvature has zero Ricci curvature and is all Weyl curvature for a sourceless region. We can however "map" the Unruh effect to the black hole case. This is done by considering the Unruh case as a case of an observer close to the horizon on an accelerated frame. The Hawking radiation emitted for large $\rho$, which persists because of spacetime curvature can be realized by the substitution $\omega~\rightarrow~tc^3/4GM$ with then gives the temperature for the black hole
$$
T~=~\frac{\hbar c^3}{8\pi k_B GM}.
$$
This is a quick way to think of Hawking radiation.
In the Penrose diagram above there is the emergence of a particle EPR pair in region I and II as marked in red. There is also a blue hyperbolic curve in regions I and II. The event horizons marking regions I and II from the black hole regions III and IV are decoupled. The two black holes are less entangled, or in a sense no longer entangled. The blue horizons exist because the red particles act as a tiny Einstein lens that reduces the size of the horizon. This "jump" in a classical setting is put in by hand as a metric back reaction. However, if we understood quantum gravity more fully we would see this as a quantum superposed system. The event horizon would in a sense be a quantum system.
There are some reasons to think this would be the case. In holography all quanta or strings that compose a black hole are on the stretched horizon just a string or Planck length above the event horizon. The strings will form up in a long Ising-like chain or $1-d$ Toda lattice, which defines a sort of quantum membrane. This quantum membrane is due to a space filling process of this long chain of strings. Also holographic principle suggests the event horizon contains a quantum field theory that is equivalent to the gravitation in the larger spacetime outside. So this horizon should play a role in the quantum physics.
This would reflect where Hawking radiation as we understand it gives way to a more fundamental understanding of nature. As yet there is not a complete picture of this.
