Technical

GenAI

Generative AI systems producing text, images, or other content.

How the laws define this

Both regulatory instruments agree on the core concept that Generative AI (GenAI) refers to AI systems that produce content, including text and images, by learning patterns from training data. Their main divergence lies in the intended context of use, the operational conditions, and the specific risks highlighted. The Coalition of 42 State Attorneys General Letter to AI Companies on Chatbot Safety defines GenAI through the lens of systems producing outputs autonomously, framing obligations and risks specifically around models that deliver human-facing content. Conversely, the Responsible AI Guidance for the Public Service: GenAI centers its definition on systems creating novel content, explicitly expanding the potential outputs to include audio or video. Furthermore, while the Attorneys General letter focuses broadly on obligations for human-facing content, the public service guidance grounds its application specifically in public service use and enumerates distinct risks, namely hallucination, misinformation, and bias.

Synthesised from the 2 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.

Jurisdiction	Instrument	Status	What is distinctive
United States	Coalition of 42 State Attorneys General Letter to AI Companies on Chatbot Safety	In Force	This definition adds emphasis on autonomous production of outputs and frames GenAI around risks associated with models that generate human-facing content from chatbots.
New Zealand	Responsible AI Guidance for the Public Service: GenAI	In Force	This definition widens outputs to explicitly include audio or video alongside text and images, and contextualizes GenAI for public service use with risks like hallucination, misinformation, and bias.

What it triggers

In the United States, the Coalition of 42 State Attorneys General Letter sets out 16 safety standards and warns against deploying harmful chatbot outputs from GenAI.source
Under New Zealand's Responsible AI Guidance for the Public Service: GenAI, the guidance describes risks associated with GenAI such as hallucination, misinformation, and bias in public service use.source

Definitions (2)

Generative AI (GenAI) refers to AI systems capable of producing text, images, or other content outputs autonomously based on learned patterns from training data. The letter uses the term to frame obligations and risks associated with models that generate human‑facing content.

Coalition of 42 State Attorneys General Letter to AI Companies on Chatbot SafetyDefinition 1 of 2

Generative AI (GenAI) is defined as systems that create novel content by learning patterns from training data and producing outputs such as text, images, audio or video; the guidance frames GenAI in the context of public service use and associated risks (hallucination, misinformation, bias).

Responsible AI Guidance for the Public Service: GenAIDefinition 2 of 2

Not to be confused with

Generative artificial intelligence

Generative artificial intelligence emphasizes producing synthetic content that emulates training data, while GenAI focuses on creating novel content from learned patterns including specific outputs and risks.

generative AI services

Generative AI services specifies systems creating novel content across text, images, audio, and video, aligning closely with GenAI but framed as services rather than general systems.

Questions and answers

What obligations does the US state AG letter impose on GenAI developers?

The Coalition of 42 State Attorneys General Letter sets out 16 safety standards and warns 13 generative AI developers against deploying harmful chatbot outputs. It frames risks for models generating human-facing content.

What risks does New Zealand guidance highlight for GenAI in public service?

The Responsible AI Guidance for the Public Service: GenAI identifies risks such as hallucination, misinformation, and bias. It guides public sector agencies on deploying GenAI.

How do these GenAI definitions differ in scope of outputs?

The US AG letter definition covers text, images, or other content outputs. The New Zealand guidance explicitly includes audio or video alongside text and images.