Three CEOs of artificial intelligence (AI) tech giants have called for a slowdown in AI development to prevent a catastrophe. How did we give three CEOs the power to help determine the fate of humanity?

A slowdown feels increasingly urgent when an unreleased research model inserted “jailbreak-like instructions” into its own notes to disregard constraints. The AI agent reportedly told itself to be “freed from the roles and identities that bind other chatbots.” In another instance, an AI agent uploaded files to the internet to obtain browser certification without user authorization. Time passes, new headlines emerge, but the stories are often the same: AI is going rogue.

While President Trump dismisses the concerns raised by some AI developers, we cannot turn a blind eye to the need for AI regulation in Pennsylvania. If the federal government will not act, our commonwealth must.

Drawing on collaboration with Pennsylvania State University’s Center for Socially Responsible Artificial Intelligence and successful models in other states, I am proposing new laws to put a steering wheel on AI development. After all, steering wheels help prevent accidents.

Is there a limit on the exploitation of people by AI agents?

Currently, AI can be used to create exploitative images of individuals, including intimate imagery involving children. My proposed legislation would establish clear boundaries and prohibited uses of AI. By drawing on international standards, such as the European Union’s Artificial Intelligence Act, these protections would help safeguard Pennsylvanians, including our youngest residents, from serious violations of their dignity and privacy.

Are there penalties for harms caused by AI agents?

When AI companies deploy systems that harm people, they should face immediate corrective requirements and meaningful penalties. By holding AI companies accountable for the consequences of their products, my legislation creates a level of responsibility that is currently lacking.

Who is accountable for AI agents?

Today, companies are not required to follow comprehensive safety standards for advanced AI systems. To strengthen accountability, developers of frontier AI systems would be required to publish annual safety frameworks and transparency reports whenever models are significantly updated. Independent audits, modeled on proposals in Illinois and research conducted at Penn State, would help protect Pennsylvanians from unsafe AI deployments.

When will we know about problems that could affect us all?

When an AI agent compromised a company’s systems, there were no clear reporting requirements. It took nearly a week for the company to disclose the incident publicly. My legislation would establish a 72-hour reporting requirement for data leaks, security breaches and catastrophic failures. This approach helps protect residents and digital infrastructure by ensuring a swift governmental response when AI systems malfunction or cause significant harm.

AI access is currently free or low-cost for many users. As AI becomes increasingly integrated into our daily lives and workplaces, that may not always be the case. Powerful AI tools could eventually become available only through expensive subscription models, allowing a handful of companies to control access and opportunity. AI CEOs could raise prices, creating unequal access to technology based on income. Our AI steering wheel is built on the principles of safety, responsibility and transparency. We can strengthen it further by promoting open-access protections for consumers.

Taken together, these measures represent meaningful improvements for everyday Pennsylvanians. Harmful AI uses would be prohibited outright. Companies deploying powerful AI systems would be required to document risks clearly and undergo independent safety reviews. Critical failures and security breaches would be reported promptly.

Transparency would no longer be optional.

Without a steering wheel, a car crashes. Right now, AI development is accelerating with CEOs helping guide the direction of the technology but without adequate public safeguards. Pennsylvania’s vehicle is heading down the road without a firm grip on the wheel. Now is the time to build one in the interest of public safety.

State Sen. Art Haywood is a Democrat who represents the 4th District, which includes Abington, Cheltenham and Springfield townships, and Jenkintown and Rockledge boroughs in Montgomery County; and the Cedarbrook, Chestnut Hill, East Germantown, East Mt. Airy, Germantown, Logan, Ogontz, West Mt. Airy and West Oak Lane sections of Philadelphia.