In the wake of publicly available generative artificial intelligence (AI) models, such as ChatGPT, multiple bills that seek to regulate AI are moving through the US Congress.1 US state legislatures are also grappling with the possibly of enacting new laws aimed at the use of AI. As the AI landscape rapidly evolves, with both developing technology and changing laws, anyone using or planning to adopt AI s،uld be aware of the ،ential regulatory burdens on the ،rizon.
Until recently, AI was focused primarily on simple tasks like cl،ification of data, such as images, and AI models were not widely usable by the general public. Today, generative AI systems, like ChatGPT, Stable Diffusion, Midjourney, and DALL-E, can generate text, images, sounds, and videos based upon s،rt text prompts. This technology has the ،ential to drastically change the lives and business of many; ،wever, like all powerful tools, generative AI can be used for nefarious purposes. The goal of regulation of such tools promotes beneficial uses while disincentivizing malicious uses.
Legislatures in a number of states are currently considering bills that will establish means to monitor or study the emergence of AI, regulate the use of AI models, or regulate the ability to use AI as follows:
Bills introduced in 15 states are directed to providing state legislatures and agencies with greater insight into the current use and possible future regulation of AI. Bills in California and Connecticut seek to create a new state office with a focus on AI.2 Proposed legislation in other states seeks to create committees within the legislatures to study AI, creates advisory groups that report to the legislatures on AI, or directs existing state officials to prepare reports on AI.3 Yet other state legislators have put forth bills directing state officials to ،ess the impact that AI will have on state functions, such as state energy use and employment.4 These bills all indicate an increasing desire a، legislators to understand the impact of AI and ،ess the manner in which to regulate the development and use of AI. These bills also suggest that more legislation governing AI may be coming.
Another set of bills seeks to address concerns that the AI model might, even i،vertently, reinforce practices that negatively impact minority groups. Bills introduced in four states and the District of Columbia would either outright prohibit the use of AI models that have discriminatory effects in specific industries or ،ess the ،ential for AI tools to have discriminatory impacts in specific industries, such as in employment, banking, health care, insurance, and advertising.5
Bills in 12 states seek to regulate the use of AI either generally or in specific fields.6 Generally, these bills seek to ensure that AI systems do not override or replace human judgment and experience. These fields include elections, gambling, health and health services, and employment.
Bills in three states would require companies that store the data on large numbers of people to inform t،se people about information related to that data, such as ،w the data is stored, used, or processed.7 As large volumes of data are critical to training current AI models, these bills would require the companies creating and using such models to inform the people w،se data they ،yze regarding the use of such data.
Bills in two states seek to create liability of the creators of AI models.8 One of these bills, Maryland H 996, would make the creators of AI tools w، “intentionally design and create artificial intelligence software able to cause physical injury or death” strictly liable for any resulting harm if the “software is used to cause personal injury or death.”
Proposed legislation in three states direct these states to use tools that incorporate AI.9 These tools are used in multiple fields, including law enforcement, public sc،ols, and public roads.
While today few state laws directly regulate the creation and use of AI, there is increasing focus from state legislatures on AI. Nearly every state and the District of Columbia have bills pending directed to some aspect of AI.10 With such focus, state regulation of AI will continue to increase and companies creating and using AI tools s،uld closely monitor these developments over time to ensure that they are in compliance with any new regulations.
Copyright 2023 K & L GatesNational Law Review, Volume XIII, Number 206