Trust and Risk-aware Human–AI Collaboration Framework in AI-Enabled OT Systems

Artificial Intelligence (AI) is increasingly integrated into Operational Technology (OT) and Cyber Physical Systems (CPS), offering enhanced automation and predictive capabilities. However, this integration introduces new risks, including loss of human control, degraded trust, and vulnerabilities to adversarial threats. Existing frameworks either focus on technical safety or human factors, and lack actionable mechanisms linking trust, control, and cybersecurity in OT environments. This paper bridges the gaps by proposing a control-theoretic framework for human-AI collaboration that is risk-aware, trust-sensitive, and aligned with formal governance standards. We illustrate the workings of the proposed framework by applying it to a realistic case study. The results provide valuable insight into Human-AI collaboration in terms of control authority and responsibility and the applicability of the proposed method in critical safety systems. The proposed framework not only manages the Human-AI collaboration during system operation but also makes sure that AI integration to the system complies with updated requirements.