In this paper, a new pseudo-random number generator (PRNG) based on chaotic\niterations is proposed. This method also combines the digits of two XORshifts\nPRNGs. The statistical properties of this new generator are improved: the\ngenerated sequences can pass all the DieHARD statistical test suite. In\naddition, this generator behaves chaotically, as defined by Devaney. This makes\nour generator suitable for cryptographic applications. An illustration in the\nfield of data hiding is presented and the robustness of the obtained data\nhiding algorithm against attacks is evaluated.\n