6G is the next generation for the communication systems. In recent years,\nmachine learning algorithms have been applied widely in various fields such as\nhealth, transportation, and the autonomous car. The predictive algorithms will\nbe used in 6G problems. With the rapid developments of deep learning\ntechniques, it is critical to take the security concern into account to apply\nthe algorithms. While machine learning offers significant advantages for 6G, AI\nmodels' security is ignored. Since it has many applications in the real world,\nsecurity is a vital part of the algorithms. This paper has proposed a\nmitigation method for adversarial attacks against proposed 6G machine learning\nmodels for the millimeter-wave (mmWave) beam prediction with adversarial\nlearning. The main idea behind adversarial attacks against machine learning\nmodels is to produce faulty results by manipulating trained deep learning\nmodels for 6G applications for mmWave beam prediction use case. We have also\npresented the adversarial learning mitigation method's performance for 6G\nsecurity in millimeter-wave beam prediction application with fast gradient sign\nmethod attack. The mean square errors of the defended model and undefended\nmodel are very close.\n