Everyday more and more devices communicate information over the Internet and the growing demand for protection has become a real challenge for civilization. While security IT systems based on conventional intrusion detection technique are simply not effective in detecting, assessing and countering cyber-threats, the use of AI based systems, thanks to their autonomy, fast paced threat analysis and decision-making capabilities, may guarantee confidentiality, integrity, and availability within the digital environment. However, increased use of AI in cyber defense may create new risks. The aim of this paper is to show experimented applications of AI for cybersecurity, raise awareness on emerging security risks that may hamper the potential of these applications in digital environment and identify possible technological solutions, best practices and legislative interventions to prevent these risks and mitigate intentional and unintentional harmful outcomes of AI based technologies.