AI & Cybersecurity

AI, or at least machine learning, and cybersecurity interact with each other. On the one hand, machine learning can be used to implement some security services (most often, reactive security services: malware classification, intrusion detection, and alert correlation). On the other hand, the increasing use of machine learning is attracting the interest of attackers who are developing new forms of attacks specifically targeting these systems: poisoning, evasion, and oracle attacks. Fighting these new forms of attacks is sometimes possible using traditional security. However, some very specific attacks that exploit the functioning of machine learning algorithms require very specific responses. This new field of study requires cooperation between security experts and AI experts. The security of the mechanisms used to provide security services is of particular concern.
In this session, experts from France, Germany, and Japan will exchange views on existing initiatives involving security experts and AI experts, and their vision on the benefits of international cooperation to win the race against hackers.