liveBook · Manning

Generative artificial intelligence

As explained in AI Governance

Generative AI is a class of artificial intelligence models that produces new content such as text, code, or images in response to inputs.

Generative AI models create content such as text, code, and images in response to inputs. Their rapid adoption and broad flexibility make governance an operational concern: organizations need to manage risks while still enabling useful applications. section 1.0 section 8.0

Unlike ordinary infrastructure, generative systems behave probabilistically, can affect people’s rights and judgment, and may let agents interpret language and chain tools. Governance makes uncertainty visible, assigns responsibility for remaining risk, embeds controls in ordinary processes, and uses failures to improve safeguards. section 8.0

Deployment risks include:

  • Plausible but fabricated outputs, such as AI-generated case law in legal filings. section 1.0
  • Instructions hidden in documents that manipulate a system’s behavior or stored memories. section 1.0
  • Exposed private chat logs and API keys, as well as prompt injection and unexpected tool-use paths. section 1.0 section 8.0

How AI Governance covers it

How deep it goes
Frames policy, technical controls, continuous improvement, and uncertainty without detailing model internals.
Examples
Cites fabricated legal cases, hidden instructions that create false memories, and exposed private chat logs.
What you can do after
Evaluate governance needs by making uncertainty visible, assigning responsibility, controlling actions, and collecting evidence.
How it connects
Connects generative AI with governance, risk management, prompt injection, and AI agents.

Also explained in

Compare how 3 books explain generative artificial intelligence →

Related concepts

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