Important AI Regulations Proposed in California: What You Need to Know

Important AI Regulations Proposed in California: What You Need to Know
Christopher Morris/ Getty Images

Warnings from industry leaders have emerged due to extensive developments in artificial intelligence, cautioning about the possibility of major risks such as rogue weapon systems and large-scale cyberattacks.

This week, a significant bill was proposed by a state legislator in California, where many of the largest AI companies are based, to introduce regulations addressing these dangers.

The bill dictates compulsory examination for far-reaching AI products prior to release. Additionally, the bill specifies that every major AI model should have a mechanism to deactivate the technology in case of malfunction.

State Sen. Scott Wiener, sponsor of the bill, emphasized the importance of establishing protective measures for extreme hazards before they occur, rather than attempting to address the risks after the fact.

Outlined below are details about the bill and its potential impact on AI regulation across the country:

How would the bill oversee AI risks?

The bill would intensify the scrutiny of large AI models before their widespread use, ensuring that state officials assess the products before they are made available to users.

In addition to an emergency off-switch, the bill would incorporate safeguards against hacking to reduce the vulnerability of AI to malicious actors.

To reinforce enforcement, the legislation would create the Frontier Model Division within the California Department of Technology responsible for implementing the regulations.

According to Wiener, the bill’s focus on extreme risks means that it will not apply to small-scale AI products. He emphasized the goal of promoting innovation while keeping safety in mind.

Furthermore, the bill would foster AI development by establishing CalCompute, a publicly owned initiative designed to facilitate shared computing power among businesses, researchers, and community groups. This effort would lower the technical barriers for smaller firms or organizations lacking the extensive computing capacity of large companies.

Sarah Myers West, managing director of AI Now Institute, commended the proactive approach of the bill in addressing and mitigating harms before they enter the market. However, she highlighted that many current risks posed by AI, such as bias in algorithms used for worker pay and healthcare access, remain unaddressed.

Wiener acknowledged that while the California legislature has introduced other bills to tackle ongoing harms caused by AI, the goal is not to solve every problem with a single bill.

Impact on AI legislation nationwide:

The California measure addressing extreme AI risk coincides with a surge of AI-related bills introduced in statehouses across the country. It is noteworthy that California’s regulations often set the standard due to the presence of many major AI companies within the state.

Despite ongoing discussions and hearings, Congress has made limited progress towards a comprehensive measure addressing AI risks. This underscores the crucial role that states play in the development of AI legislation.

Dylan Hoffman, executive director for California and the Southwest at industry lobbying group TechNet, emphasized the significance of U.S.-based AI regulation in shaping global rules related to the technology. He expressed optimism about the potential of the legislation to benefit all Californians, mitigate risks, and enhance global competitiveness.

Wiener mentioned that while the bill borrowed some concepts from an executive order issued by President Joe Biden in October, he remains skeptical about the prospects for federal legislation mirroring the approach taken by the California bill.

Copyright © 2024, ABC Audio. All rights reserved.

Leave a Reply

Your email address will not be published. Required fields are marked *