Ai risk management

As explained in Introduction to Generative AI, Second Edition

A risk-management system is an organized process for identifying, assessing, controlling, and monitoring risks associated with an AI system.

AI risk management organizes the identification, assessment, control, and monitoring of risks associated with an AI system across its lifecycle. The European Union’s AI Act applies stronger obligations where systems may affect safety or fundamental rights. section 10.30

The AI Act distinguishes four broad risk levels:

• High-risk systems, including AI used in hiring, law enforcement, education access, and financial-services access, require assessment before market entry and throughout the lifecycle. section 10.30
• Limited-risk systems must meet transparency requirements, such as informing users when they interact with chatbots, deepfakes, or emotion-recognition systems. section 10.30
• Minimal-risk applications, such as video games and spam filters, are mainly addressed through voluntary codes of conduct. section 10.30

High-risk systems must support data governance, accuracy, nondiscrimination, record-keeping, technical documentation, human oversight, and registration in an EU-wide database. Noncompliance can result in large fines. section 10.30

Figure 10.7 The four risk categories in the AI Act

A visual summary of the AI Act’s four risk categories helps compare regulatory intensity across system types. section 10.30

How Introduction to Generative AI, Second Edition covers it

How deep it goes

Identifies risk tiers and required controls but does not explain how to implement risk assessments or monitoring procedures.

Examples

Uses hiring, law enforcement, education access, financial services, chatbots, deepfakes, video games, and spam filters as examples.

What you can do after

Evaluate AI applications by risk level and identify corresponding oversight, documentation, transparency, and governance duties.

How it connects

Connects risk management with human oversight, data governance, transparency, technical documentation, and nondiscrimination.

liveBookmanning.com

Read Introduction to Generative AI, Second Edition on liveBook Get the book at manning.com

Related concepts

• AI-specific governancepart of · AI governance is the set of policies, processes, controls, and oversight mechanisms for directing and managing the development and use of artificial-intelligence systems.
• red-teaminguses · Red-teaming is an adversarial assessment practice that deliberately simulates attacks or uses challenging inputs to identify a system’s vulnerabilities, failures, or unsafe behavior.
• threat modeluses · A threat model is a structured analysis or representation of a system’s relevant threats, adversaries, assets, attack paths, assumptions, and possible mitigations.
• impact assessmenthas part · An impact assessment is a systematic evaluation of a system’s potential effects, harms, or risks for affected people or society.

This explainer was generated by Manning's knowledge graph from the text of Introduction to Generative AI, Second Edition. The section links point to the parts of the book it draws on.