Zero Trust Architectures for Responsible AI in Enterprise Applications

Abstract

The current-day business decisions have been centered on artificial intelligence (AI). Nonetheless, obtaining them without compromising transparency and ethics is becoming a problem. The paper has explored how the application of Zero Trust Architecture (ZTA) and the principles of Responsible AI (RAI) can be used to build AI systems that are both secure and ethically right. We research existing frameworks, field testing, and design techniques that deploy ZTA to the AI lifecycle - collecting data to training, deployment, use, monitoring, and auditing. We have found that ZTA does not only make security to be more secure through hard access control and privacy, but it also increases the level of fairness, transparency, and resistance to adversarial attacks. We finish by a couple of suggestions to future research on ZTA-powered Responsible AI and conclude with the argument that one model of governance can unite both cybersecurity and AI ethics.