The Generative AI Regulations Are Coming — Current State Movements on Regulating Generative AI
As with any disruptive technology, such as Uber and facial recognition, emerging technologies, and new business models can challenge existing government rules and regulations. This often creates a fluctuating environment as new regulations emerge to try and keep pace. Generative AI, the domain responsible for creating viral new AI-driven capabilities such as ChatGPT, is currently under scrutiny by the federal government and numerous state legislatures. This activity is an extension of other AI-centric regulations that have emerged around the governmental use of automated or algorithmic decision-making. However, there is a growing focus on this new domain as many consumer and business applications are rapidly embedding generative AI capabilities in their products.
With the help of our friends at Quorum, I’ve compiled a snapshot of some emerging bills that have singled out generative AI use cases. I will continue to track this landscape over the coming months, so please send any regulation, policy, or government generative use-cases to me for inclusion in future curations.
Status — Re-referred to Committee on Government Organization (4/10/23)
The California AI-ware Act establishes the Office of Artificial Intelligence within the Department of Technology to guide the design, use, and deployment of AI systems by state agencies, ensuring compliance with privacy and civil liberties laws while minimizing bias and promoting equitable outcomes for all Californians. State agencies using generative AI to communicate with individuals must provide clear notice of AI involvement and offer instructions on how to communicate directly with a human representative. The legislation emphasizes principles of fairness, transparency, and accountability while prohibiting discrimination based on protected characteristics in AI system design and use.
(California) SB-721 California Interagency AI Working Group.
Status — Set for Hearing (4/11/23)
SB 721 introduces the California Interagency AI Working Group, which aims to deliver a report to the Legislature concerning artificial intelligence (AI). The working group members must have expertise in areas such as computer science, AI, technology, workforce development, and data privacy. The group is tasked with recommending a definition of AI for legislative use, studying AI’s implications for data collection, taking proactive steps to prevent AI-assisted misinformation campaigns, determining relevant agencies to develop and oversee AI policy, and collaborating with the California Privacy Protection Agency for data privacy policies. The working group will submit a report to the Legislature by January 1, 2025, and every two years thereafter.
(Illinois) H.B. 3563 — An Act Establishing the DOIT — AI Task Force
Status: Referred to Senate Committee on 3–23–23
The act establishes the Generative AI and Natural Language Processing Task Force under the Department of Innovation and Technology in the State of Illinois. The Task Force is composed of various members, including experts in artificial intelligence, cybersecurity, and education. It is responsible for investigating generative artificial intelligence software and natural language processing software, and providing recommendations for legislation, regulations, and policies regarding AI use in schools, public services, civil rights, workforce, and cybersecurity. The Task Force will hold public meetings and submit a report to the Governor and the General Assembly outlining their findings and recommendations.
Status: Referred to the Committee on Advanced Information Technology (2/16/23)
The bill seeks to regulate large-scale generative artificial intelligence models like ChatGPT in Massachusetts to ensure public safety, privacy, and intellectual property rights. Companies operating such models must adhere to operating standards prohibiting discrimination, implement anti-plagiarism measures, protect user data, obtain informed consent, delete unnecessary data, and conduct risk assessments. Companies must register with the Attorney General, who will maintain a public registry and enforce the bill’s provisions.
🤖AI-driven Summary
So, what are the key themes across all of the proposed legislation? With the help of generative AI — specifically, ChatGPT (GPT-4) — I’ve extracted the text from all of the bills and have summarized key aspects of these multi-state developments. From this analysis, we can see that states are moving forward on generative AI regulation by initially focusing on:
- Establishing oversight bodies: Creating specialized committees, task forces, or offices to guide and monitor AI development, deployment, and use within state agencies.
- Ensuring fairness and preventing discrimination: Prohibiting AI systems from discriminating based on protected characteristics and promoting equitable outcomes.
- Protecting privacy and data security: Implementing measures to safeguard user data, obtaining informed consent, and deleting or de-identifying unnecessary data.
- Enhancing transparency and accountability: Mandating clear notice of AI involvement in public interfaces, promoting anti-plagiarism measures, and requiring regular risk assessments.
- Encouraging interagency collaboration: Facilitating cooperation between different government bodies, industry experts, and stakeholders to develop comprehensive AI policies and best practices.
Also posted on LinkedIn.com.
