The ad begins with a woman standing in a department store who sort of looks like Gov. Janet Mills, but not quite.

“Introducing the Janet Mills collection, featuring a confusing choice that forces girls to compete against biological males,” the female narrator says over banal instrumental music as the video cuts to “Mills” holding a stopwatch by an outdoor track.

The Mills collection comes “with a no-parent-permission-required estrogen kit,” the narrator continues, as the imposter holds a kit of syringes while patting a boy’s hair, which seems suspiciously stiff. The commercial ends with a real picture of the governor.

As far as ads generated by artificial intelligence go, the one from the National Republican Senatorial Committee is not very convincing. But the commercial serves as a reminder about how the emerging technology is being integrated into political campaigns and other areas of life in Maine.

If state Democratic leaders get their way, AI-generated ads like this won’t be allowed in Maine without a disclaimer.

As AI technology rapidly improves, state policymakers are weighing a variety of measures that could affect how Mainers interact with it. They are taking a two-pronged approach to protect people, especially children, from potential harms — while also preparing for the possible benefits.

The technology comes in the form of virtual personal assistants, internet search results and targeted advertising by businesses. It’s being used by governments for things ranging from traffic signals to budgets and policymaking to facial recognition to surveillance.

Mills said in a written statement that AI could help improve lives, drive economic growth and solve complex problems, but that it must be used in a “prudent, responsible, and ethical manner.”

“As AI becomes more prevalent in our society, its considerable promise must be balanced against harms — known and unforeseen — that can emerge from its widespread use,” she said. “It’s clear we’re only at the beginning of AI’s evolution.”

The governor has proposed $6.7 million in her supplemental budget to begin implementing some of the recommendations of a 21-member task force she created last year to study the issue.

Her proposal, which is being reviewed by lawmakers, would create a statewide AI literacy campaign; fund local and state partnerships to help municipalities use the technology and offer grants to support job training programs to keep Maine’s workforce competitive and productive in AI-enabled workplaces, among other things.

Lawmakers, meanwhile, are considering bills to address potential harms. In a rare bipartisan move, Republicans and Democrats voted unanimously last month in support of a bill (LD 524) making AI-generated child sex abuse material illegal. But that bill must receive about $55,000 before it can be sent to the governor.

They are also considering bills:

• To require political ads in state and local elections to include a disclosure when AI-generated or altered material is used (LD 517).
• To stop human-like chatbots or social AI companions from interacting with children (LD 2162).
• And to regulate how the technology is used in mental health settings (LD 2082).

Last year, lawmakers passed a measure including AI-generated images in the state’s ban on so-called “revenge porn,” and one requiring companies to inform consumers when they’re interacting with an AI assistant. Mills signed both into law.

Other proposals regulating AI use in medical and dental insurance claims and in setting rents died in committees. So did one prohibiting the use of AI in “dynamic pricing,” in which businesses use the technology to offer different real-time prices to different consumers.

Over 1,000 measures focusing on artificial intelligence were debated in state capitols last year, the National Conference of State Legislatures said.

Some states, such as Colorado and California, are taking steps to enact a broad regulatory framework for AI. California has provisions preventing discrimination in the workplace and requiring watermarks on AI content and transparency around data used to produce reports.

But Maine lawmakers are seeking to address potential harms on a case-by-case basis — at least for now.

“I think of it as almost a whack-a-mole type of approach where we are developing legislation that very narrowly addresses specific harms of AI,” said Rep. Amy Kuhn, D-Falmouth, who is taking the lead for House Democrats.

“That sort of overarching regulatory framework just feels a little premature for Maine to me right now. I want to see that work its way through the states and let some other states take a swing before we get in there.”

Republicans, however, are worried about overregulation.

Rep. Jennifer Poirier, R-Skowhegan, said her caucus is focused on protecting children from potential harms associated with AI, but she worries that regulation will never keep up with AI’s evolution.

“You can’t always legislate your way out of everything,” Poirier said. “If you have a minor that has access to AI, and it can be used to harm them in any way, it’s our responsibility as adults to keep them safe. … But we are adults, and we need to use our own common sense.”

A recent poll from Pan Atlantic Research showed widespread concern about AI, with 66% of the 810 Mainers surveyed saying they’re mostly concerned about the potential problems of AI, while 25% were mostly optimistic.

More advanced programs can generate text, analyze reports and create increasingly lifelike images and videos. A recent AI video purporting to show Tom Cruise and Brad Pitt throwing down over the death of convicted sex offender Jeffrey Epstein raised alarm bells in Hollywood over its realism.

Other programs have been used by businesses in ways critics say can be exploitative. Consumer Reports recently published a story about how the online grocery shopping service Instacart used AI to charge certain customers higher prices — up to 23% more — if they were flagged as having higher incomes. Instacart reportedly stopped offering stores this option for “surveillance pricing” after the story was published in December.

A lobbying effort is underway to promote AI regulation in Maine. The “Protect What’s Human” campaign launched a website earlier this year, and a spokesperson said they have invested about $210,000 in ads supporting AI regulations. The commercials are targeting Republicans voters in the Bangor and Portland regions. The group is planning to spend another $110,000 on TV, streaming services, social media and podcasts.

Other proposals passed by the Legislature reflect lawmakers’ attempts to get ahead of the AI issue in indirect ways.

The House and Senate have each recently passed a strict data privacy law that would greatly restrict the amount of data — a person’s location, browsing and shopping histories and biometric information, for example— that companies can collect, store and sell. One of the main arguments was that such data can be used to train AI models. However, the chambers will have to iron out the differences between their two versions of the measure, LD 1822, if it is to become law.

And local residents are beginning to grapple with proposed data centers, which have been controversial in other parts of the county because they consume large amounts of water. This is especially true for centers powering AI.

Lawmakers are considering a bill, LD 307, to create a moratorium on such centers and establish a state council to study and review the impact of building them in Maine.

Construction is underway on a data center in Aroostook County, while another is being proposed in Sanford. Others have been proposed in Wiscasset and Lewiston, but did not move forward.

“This whole world is shifting to computer everything,” Poirier said, “and it’s important that we keep up with the times on that.”