Abstarct: Today, especially in cryptography, random numbers generation is of tremendous importance. To generate random numbers depending on their applications we use different generation methods. Security is an important parameter in cryptography. In this scientific reaserch we're attempting to generate random numbers that are not detected by eavesdropper and contains a strong security. In Newtonian physics, as the complete knowledge of initial conditions along with interactions of a system allows one to predict its future dynamics deterministically There is no securit in determinestic classical methods of generating random numbers. Quantum theory also incorporates a form of randomness in its frxamework that does not have a classical counterpart. Quantum physics is inherently random, consequently it is convenient to use random numbers generating. In this scientific reaserch we determine randomness through outputs obtained by a bell experiment, then we choose the best system in order to random number generation. According to quantum theory, the outcomes obtained by measuring an entangled state necessarily exhibit some randomness if they violate a Bell inequality. We are going to investigate the relation between non-locality and the amount of randomness necessarily present in a Bell experiment. Naively, we expect a direct relation between the amount of nonlocallity and the randomness produced in a Bell-type experiment, i.e.,for eaxample the less nonlocality, the less randomness. Our analysis, however, show that this intuition is not correct and that the relation between these two concepts is much subtler than expected. We can consider the amount of violation of the CHSH inequality a natural measure of non-locality then compute the amount of randomness in system. in this scientific reaserch We introduce little non-local correlations that violate arbitrarily little the CHSH inequality yet which necessarily imply that the maximal amounts of randomness. We know that by performing measurements with binary outcomes on two subsystems one could in principle generate up to two bits of randomness. We show that states with arbitrarily little entanglement can be used to certify that close to the maximum of two bits of randomness are produced.