Cyberbullying-related harassment is a significant issue on social media. At least one of the three bottlenecks listed below exists in the methods that are currently being used to identify cyberbullying. For the first time, they concentrate on a single social media platform (SMP). Second, they exclusively cover one theme: cyberbullying. Third, they rely upon all around made information qualities. We demonstrate that these three obstacles can be overcome by deep learning models. It is possible for these models to transfer their knowledge to new datasets. In our broad preliminaries, we utilized three genuine world datasets: Twitter (16k posts), Formspring (12k posts), and Wikipedia (100k posts). On how to spot cyberbullying, our research sheds light on a number of important points. Supposedly, this is the principal examination to completely inspect the recognition of cyberbullying across various subjects.