AI systems are increasingly being used to find and exploit vulnerabilities. A fully fledged use of AI for cybercrime would render human cybersecurity useless, as such attacks could be performed at super-human speeds by algorithms that constantly learn how to dodge security measures, making such attacks much more effective and far-reaching. In addition, AI can be used for social engineering (fool humans with voice cloning, deep fakes, etc.) and gain easy access to otherwise secure systems; to break certain biometric security systems so that, for example, stolen accounts aren’t flagged by behavioural security systems; or it can be embedded to malware to search for specific information.