A Layered overview of Threats and controls to Mitigate harmful Generative AI solutions
When developing and using Generative AI solutions, as a Security professional, you will need to understand what are the threats that the technology might be exposed to.
Understanding attack surface is key to our jobs , of course.
Let’s discuss this.
Generative AI Layers
While I’m not a Data Scientist nor an AI Architect, over the last year I have been exposed to a range of sessions and training material at Microsoft to understand enough about Generative AI technologies to discuss the subject.
Particularly because of the launch of Copilot for Security, having these deep, architectural discussions is something customers adopting the new technology are curious and mindful of.
Therefore, why this research.
When we look at Genetive AI solutions’ Architectural blocks and potential threats to services, we can talk about 5 layers, essentially.
If you want details on each one of them, please check out my video: