By looking to the circuit for the QFT presented in the M. Nielsen and I. Chuang textbook (Figure 5.1.) we can notice that all controlled rotations can be neglected because for each control rotation gate the control qubits are in $| 0\rangle$ state (for the case described in the question). Here is the Figure from the book:
So effectively, in this case, we have only Hadamard gates acting on $n$ qubits. And, of course, Swap gates (will talk about it later). So
$$H^{\otimes n} |000...0\rangle = \frac{1}{\sqrt{2^n}} \sum_x{|x\rangle}$$
where $H^{\otimes n}$ is the notation of the $n$ Hadamard gates acting on $n$ qubits that are all in the $|0\rangle$ state, $\sum_x{|x\rangle}$ is the sum of all bit strings (from $|000...0\rangle$ to $|111...1\rangle$). Swap gates at the end of the QFT circuit (not presented in the Figure 5.1.) will not change this state: the final state after the swaps will stay the same. So all bit strings have equal amplitudes $\frac{1}{\sqrt{2^n}}$, and therefore equal probabilities $\left|\frac{1}{\sqrt{2^n}}\right|^2 = \frac{1}{2^n}$. Thus, the answer is: for 10 qubits we will measure all qubits in the $|0\rangle$ state with $\frac{1}{2^{10}} \approx 0.001$ probability.