Rep. Lori Trahan is revising her artificial intelligence safety bill to further narrow provisions that would override state laws governing the industry, according to a person familiar with her efforts.

The Massachusetts Democrat’s work marks another adjustment to legislation she drafted with Rep. Jay Obernolte, a California Republican. The person, who was granted anonymity to discuss private legislative work, said Trahan is revamping the bill’s preemption language.

Trahan and Obernolte released their original proposal in June. Its provisions limiting state regulation drew criticism from AI safety groups, prompting the lawmakers to substantially narrow that language when they introduced the bill in August. Trahan’s latest revision would further restrict the scope of preemption, according to the person familiar with the effort.

The issue has exposed a divide over how federal and state governments should share responsibility for overseeing artificial intelligence. Democrats largely oppose preventing states from creating their own AI regulations. The technology industry, meanwhile, argues that navigating a patchwork of different state laws is untenable.

Republicans generally favor preemption, but their influence over the legislation will be limited next Congress, when Democrats take control of both chambers.

The revisions reflect the challenge of crafting a federal AI safety measure amid competing views about the role states should play. The initial response from safety groups led to narrower language before the bill’s August introduction; Trahan is now seeking to narrow it further.