Most organisations now have AI principles. Far fewer have changed how they actually build and deploy systems. Responsible AI in practice is the work of closing that gap: deciding who is accountable, assessing risks before a system is released, testing it for failures and harms, documenting what it can and cannot do, and monitoring it afterwards. Everything earlier in this course, from bias and privacy to safety and regulation, shows up here as concrete decisions made by real teams.
The lessons begin with an introduction to the NIST AI Risk Management Framework, a widely used voluntary framework for governing, mapping, measuring and managing AI risk. A short explainer on AI red teaming then looks at how systems are deliberately stress-tested to find failures before real users do. The final lesson is a lecture by a researcher who worked on the NIST framework, reflecting on what a voluntary approach can and cannot achieve.
Practices and standards in this area are still developing, and frameworks only help if organisations act on them, so treat these lessons as an introduction rather than a checklist. When you have finished, continue on to the course Projects to apply what you have learned.
A walk-through of the NIST AI RMF and its four functions, govern, map, measure and manage, for developers, business leaders and compliance teams.
Video 1minA short explainer from a cybersecurity company on red teaming: simulating attacks on chatbots, agents and AI-powered workflows to find weaknesses before real users do. Focuses on security testing.
Video 2minJeanna Matthews, who worked on the framework at NIST, gives a detailed overview of how it was developed and discusses the implications of its voluntary nature. A long, advanced, optional watch.
Video 1h 2min