Artificial intelligence is continuously used in both the public and private sectors with an increasing tendency. This is further driven in particular by the availability of so-called low/no code development frameworks. However, increased use combined with sometimes little understanding due to the black-box nature results in increased requirements, especially for security and privacy. Measures that address the aforementioned requirements and contribute to the development of secure software are already well established in traditional software development, but not in AI development.
Complicating matters is the fact that development options such as no/low and high code in turn harbor quite different security requirements. The goal is to address the different development variants and to highlight special security requirements, as well as to discuss their differences.