Generative AI (GenAI)
AI that produces new content like text, images or code.
How the laws define this
Regulators consistently define generative AI as systems or models capable of producing new or novel content—such as text, images, audio, video, or code—by learning patterns from existing training data, typically in response to user prompts or inputs. Where these instruments diverge is in how they explicitly set boundaries around this capability relative to other technologies. The Guidelines for the Responsible Use of AI in the Public Service distinguishes generative AI from targeted or narrow systems, whereas the Interim Generative AI guidance for the public service explicitly contrasts it with tools that only classify or predict. Taking a different approach, the Selangor AI White Paper sets its boundary around human mimicry, requiring that outputs closely resemble human-created content. Finally, the National Cyber Security Centre: Cyber Security Guidance on Generative AI for Public Sector Bodies uniquely incorporates concrete technological examples directly into its definition, specifically scoping the term to encompass chatbots, large language models, and diffusion models.
Synthesised from the 9 statutory definitions below. Each one is quoted in full, with its source.
How jurisdictions differ
What is distinctive in each definition, binding law in force first. Showing 7 of the 9 instruments that define this term; all are listed under Definitions below.
| Jurisdiction | Instrument | Status | What is distinctive |
|---|---|---|---|
| California, United States | As Trump rolls back protections, Governor Newsom signs first-of-its-kind executive order to strengthen AI protections and responsible use | In Force | Includes audio and video alongside text and images in the list of generated content types. |
| Georgia, United States | Georgia State AI Governance and Use Guidance | In Force (Amended) | Emphasizes that systems create new and original content by learning patterns from existing data. |
| Ireland | Guidelines for the Responsible Use of AI in the Public Service | In Force | Distinguishes generative AI from targeted or narrow systems and highlights risks like data leakage. |
| New Zealand | Responsible AI Guidance for the Public Service: GenAI | In Force | Specifies that systems generate outputs in response to inputs or prompts. |
| British Columbia, Canada | Draft Artificial Intelligence Responsible Use Principles | Draft | Provides examples of applications like summarizing documents, answering questions, editing text, and producing creative content. |
| Malaysia | Selangor AI White Paper | Adopted | Highlights that output closely mimics human-created content in creative and data synthesis tasks. |
| Ireland | National Cyber Security Centre: Cyber Security Guidance on Generative AI for Public Sector Bodies | Superseded | Lists specific examples like chatbots, large language models, and diffusion models, focusing on cybersecurity and privacy risks. |
What it triggers
- In California, state procurement requires safeguards against data exploitation, bias, and civil rights violations.
- Georgia's executive branch agencies follow security and ethical standards including prior AI approval.source
- Ireland's public service guidelines prescribe tailored end-user controls and restrictions on uncontrolled use of free public GenAI services.source
- New Zealand's public sector agencies address specific governance, privacy, and risk considerations.source
Definitions (9)
AI systems that generate novel outputs (e.g., text, images, audio, code) based on learned patterns, distinguished in the Guidelines from targeted or narrow systems; the guidance highlights specific risks such as data leakage and inappropriate use and prescribes tailored end‑user controls and restrictions on uncontrolled use of free public GenAI services.
AI systems that generate novel content—such as text, images, audio or code—by learning patterns from training data, as distinct from systems that only classify or predict; the document uses this operational definition to scope guidance and risk management.
Systems capable of producing new content — text, images, audio or video — by learning patterns from training data and generating outputs in response to inputs or prompts. The guidance uses this term to distinguish models used to synthesise content from other AI systems and to frame specific governance, privacy and risk considerations.
Systems and models capable of producing new content such as text, images, audio, video or code from prompts or inputs; examples include chatbots, large language models and diffusion models. The guidance treats GenAI as the class of technologies whose outputs and training/data handling create specific cybersecurity and privacy risks.
AI models capable of generating new content (text, images, audio, video, or other media) by learning patterns and structures from training data, producing novel outputs with characteristics similar to the input data.
Generative AI (GenAI) refers to advanced AI tools or systems that can create new and original content, including text, images, video, or audio. These systems learn patterns from existing data to generate novel outputs.
Generative AI (Gen AI) describes any form of AI that is capable of creating new content, which can include text, images, video, or audio. These tools operate by learning patterns and structures from vast datasets and then generating novel outputs in response to user prompts. Examples of Gen AI applications range from summarizing documents and answering questions to editing text and producing creative content.
Generative AI (Gen AI) specifically denotes AI systems capable of producing novel content such as text, images, or other data. This output closely mimics human-created content, showcasing advanced capabilities in creative and data synthesis tasks.
A specific type of Artificial Intelligence capable of creating new content, such as images, text, audio, or video, often based on patterns learned from existing data.
Not to be confused with
Refers to advanced systems with general-purpose generative capabilities, while this term covers all generative AI without specifying advancement level.
Uses 'IAG' abbreviation but defines similarly as systems generating novel content like text, images, audio, or code.
Questions and answers
What safeguards are required for generative AI in California state procurement?
California's Executive Order N-5-26 requires safeguards against data exploitation, bias, and civil rights violations. It strengthens AI protections and responsible use.
What risks does Ireland highlight for generative AI in public service?
Ireland's guidelines highlight risks such as data leakage and inappropriate use. They prescribe tailored end-user controls and restrictions on uncontrolled use of free public GenAI services.
How does New Zealand distinguish generative AI in its public service guidance?
New Zealand defines it as systems producing new content from training data in response to prompts. The guidance frames specific governance, privacy, and risk considerations to distinguish it from other AI systems.
Related Terms
Generative Artificial Intelligence (GAI)
Models that produce novel content from learned patterns....
Advanced generative AI systems
AI systems with general-purpose generative capabilities....
Generative AI (IAG)
Systems generating novel content (text, images, audio, code)....
Tools based on generative artificial intelligence
Systems that autonomously generate content using deep‑learning models....
Generative Artificial Intelligence (GenAI)
AI systems that generate new content (text, images, audio, etc.)....