Gov. Ned Lamont and fellow Democrats could be headed for a collision course over artificial intelligence – for a second straight year.

Calling AI the “Wild West,” the Connecticut Senate approved sweeping new regulations late Wednesday night.

After supporters stripped out the most controversial sections, the bill passed 32-4. But the changes may not be enough to avoid a veto from Lamont.

Artificial intelligence is just about everywhere. It can speed up Amazon deliveries, improve your internet searches and lower your doctor’s bill. ChatGPT can even write a term paper for you.

But AI can also be used for scams and “deep fakes”, where highly-sophisticated software impersonates you or someone else. It can also make automated, life-altering decisions – like denying you health coverage.

Now, some state lawmakers want to regulate the rapidly growing technology.

The bill now headed to the Connecticut House of Representatives requires that consumers be notified when AI is being used, what data it’s using and what decisions it’s making.

“We’ve seen the Trump administration literally put regulations in the shredder that President Biden worked on with stakeholders,” said Senate Majority Leader Bob Duff (D-Norwalk). “Now, it’s literally the ‘Wild West.’”

Consumers could also appeal decisions made by AI if the determination was based on materially incorrect information.

But the legislation faces an uphill battle. Before Thursday’s debate, Lamont said the law could scare away new tech startups.

“I just worry about every state going out, doing their own thing, a patchwork quilt of regulations. Connecticut being probably stricter and broader than most,” he told reporters.

All day on Wednesday, top Senate Democrats huddled to reach a compromise with Republicans.

They removed a ban on "algorithmic discrimination" – including discriminating against people applying for jobs, loans, health care, college admissions and housing.

State Sen. Gary Winfield (D-New Haven) said the watered-down bill will do little to protect minorities.

“I’m voting for this bill," he said. "But let's not kid ourselves. This bill has a glaring hole in it.”

Lamont believes existing laws are adequate.

"I understand people's concerns about AI. I know about ‘deep fakes,'" he said. "I also know that our consumer protections laws are really strong.”

The governor supports parts of the bill that criminalize “deep fake” pornography and expanding AI research and job training.

Last year, the Senate passed similar legislation – but the Connecticut House never called it for a vote after Lamont threatened a veto.

Without his blessing, the latest version is likely to face a similar fate.

“Did they reach a deal?” said House Speaker Matt Ritter (D-Hartford). “If they were able to get there with the administration, that’s great. We’d love to run the bill.”

But some Republicans said Congress – not a small state like Connecticut – should regulate the burgeoning industry.

“We’re seeing in California and Colorado – the laws that they’re putting forth aren’t getting off the ground because they recognize one state can’t regulate this arena,” said House GOP leader Vin Candelora (R-North Branford).

The clock is ticking for both sides to reach a compromise. The General Assembly session ends in three weeks.