Safety, Ethics & Governance · beginner · concept 176 of 187

⚖️Responsible AI

Responsible AI is the operational practice of building and running AI systems against stated commitments: fairness, transparency, accountability, privacy, and human oversight. It works through process, documentation such as model cards, impact assessments, sign-off gates, monitoring, not through any single technique. NIST's AI Risk Management Framework organizes it as four functions: Govern, Map, Measure, Manage. Practitioners conflate it with AI safety, which targets harm from model capability, and with regulation, which is law. Responsible AI is voluntary, so it holds only where governance can actually stop a launch.