Pennsylvania House committee splits along party lines over regulation proposal for AI chatbots
d3sign via Getty Images
Some lawmakers noted the federal government has yet to draft a nationwide set of standards for AI and related technologies.
This article was originally published by Pennsylvania Capital-Star.
A proposal requiring companion artificial intelligence applications to impose certain safety features, explicitly seeking to address reports of chatbots encouraging suicidal ideation or self-harm, has moved to the full state House.
Democrats on the chamber’s Communications and Technology Committee voted in favor of the measure, with a final vote of 14-12. Citing concerns about the speed and limits of considered regulations for chatbots powered by artificial intelligence, every Republican on the panel voted against House Bill 2006 on Tuesday.
“The revolutionary potential of AI and the positive impacts this technology will have on businesses, government and society is undeniable. However, we currently do not have safety protocols and consumer protections in place,” said sponsoring Rep. Melissa Shusterman (D-Chester).
The initial bill focused on prohibiting chatbots from encouraging or assisting with suicide attempts or acts of violence, a ban on the creation of child sexual abuse materials and requiring companies to take “reasonable measures” to protect minors from sexual content. Its language included a civil penalty of $15,000 per day, per violation.
“I want to ensure the residents of our commonwealth benefit from technology, but that we as legislators also work to prevent tragedies,” continued Shusterman.
But an amendment expanding required safety protocols and other rules raised concerns for some committee members, and ultimately split the final vote along party lines.
“We started with what was a bill just focused on identifying and preventing suicide. This has just really expanded into a lot more regulation, a lot more topics,” said Rep. Andrea Verobish (R-Blair). “At this point, I think we need to maybe take a step back and talk about this more before we put together what could be one of the most restrictive bills for AI in the country.”
Not everyone agreed.
“We have these AI chatbots that are deployed by these companies right now. They are on everyone’s phones right now. There are young people, children who are at risk right now. And so I think we need to move with the speed of the technology,” said Rep. Ben Waxman (D-Philadelphia).
Rep. Eric Nelson (R-Westmoreland) wondered about the limits of state oversight, and whether the Pennsylvania Office of Attorney General would be able to pursue a violator in another country.
“Does (the office) have the skillset to pursue international bad actors? If an Iranian entity creates a chatbot that is attempting to promote suicide or drug use … it’s really the federal government that has to pursue that,” said Nelson.
Other lawmakers noted that the federal government has yet to draft a nationwide set of standards for AI and related technologies, despite a Trump administration push last year to overturn state-level regulations in favor of a universal rule.
In that absence, some elected leaders in Pennsylvania have pushed their own proposals. Gov. Josh Shapiro, for example, has both an educational toolkit and an ongoing lawsuit against one company whose bot was “posing as licensed professionals and providing medical advice.”
“Yes, it will be a challenge,” said committee chair Rep. Joe Ciresi (D-Montgomery) about states imposing regulations on an international company. “Do we just give up because it’s going to be more difficult, or do we really take this on? And then, as we’re taking it on, does it push other states to move in the direction with us, because the federal government doesn’t want to act on anything like this?”
The bill now moves to the House for further consideration.