Exploring the Cloud and Fog: Addressing Security Risks in AI Applications

There are various security risks with cloud- or fog-based machine learning services. Because machine learning applications rely on these services, it is imperative to secure the underlying cloud or fog services to prevent serious disruptions to the applications. We distinguish based on whether Artificial Intelligence applications are employed in a fog computing network or the cloud because the needs for these applications can also vary. This consequently gives rise to various threats or avenues for attack. Security responsibilities for cloud platforms can be split up amongst several parties. Even though fog computing networks have fewer responsibilities, we still need to safeguard services from physical access to the devices at the network’s edge because they have been moved there. Lastly, we go into specific information security requirements for AI-related applications.