A new reporter is working at Florida’s largest daily newspaper, one who doesn’t need a desk or health insurance.

The Tampa Bay Times AI News Engine has been covering beats like high-end real estate transactions and breaking news about extreme weather since September.

Readers who don’t check bylines may not have noticed. There was no dedicated press release, and there’s no article or announcement on the website. But back in August, the paper disclosed its upcoming AI bot to readers of the subscriber-only “Best of Times” newsletter. There, it’s buried under a story asking readers to purchase a Bucs anniversary coffee table book.

“You may not have noticed because it represents a tiny percentage of the content we produce on our website,” Times Executive Editor Mark Katches told Creative Loafing Tampa Bay about his new worker.

According to Katches, the Times AI News Engine is the creation of a Swiss-based third party vendor called United Robots, and it works somewhat like a Mad Lib word game. The staff creates pre-written article templates before the AI bot “writes” the stories by filling in the gaps via publicly available information scraped from government websites.

“This is additive, supplementary coverage,” explained Katches. “Coincidentally, our newsroom has actually grown in size slightly since we launched this use of AI. I say that only to drive home the point that the goal of this is not to cut newsroom jobs.”

Of course, the Times does have living, breathing reporters who cover real estate, breaking news and environmental issues, but Katches claims the bot will only free up time and “not replace any work done by the Times.”

The bot has been busy over the last six months—churning out at least 50 news stories across the Times website. For instance, last week the AI bot published a 126-word story about the sale of a $6.9 million home in Tampa’s Davis Islands neighborhood.

“A single-family home in Tampa sold on Jan. 29 for $6.49 million. It features 6,054 square feet of living space,” wrote the bot. “The home at 107 Martinique Avenue includes five bedrooms and seven bathrooms. It was sold by ECH Capital LLC to Alise G. Bartley, David W. Bartley II and the Alise G. Bartley Revocable Trust.”

The article is incredibly brief, and honestly, pretty hard to screw up. The only photo is a Google Streetview image of the property’s address, and the copy doesn’t go into any history or detail about the home, other than the price per square foot and the sales of other nearby properties.

“Every real estate story is reviewed and edited before publication,” said Katches to CL. “Again, I’ll stress, the Times wrote every template with specific blanks to be filled in when AI scrapes official sites of real estate transactions in the communities where we’ve directed it to look. These are pretty short, straightforward stories. And all these real estate stories that are generated get edited by our staff.”

Katches is right, the bot needs human oversight, as evidenced by headlines like, “How much were the top 10 most expensive home sales in Pasco, reported last week?”, a collection of words that would break any human editor’s brain.

But while the safeguards of human intervention exist on the Times’ real estate stories, the emergency weather alerts are completely left to the discretion of the AI bot.

“These may auto publish in the interest of public safety,” Katches said. “This way, when an extreme weather alert is issued–like a tornado warning or a flash flood warning that has a limited time of an hour or so, we can get these alerts posted immediately, including when we’re not staffed to edit a story.”

If the National Weather Service (NWS) posts at 3 a.m., the bot will regurgitate that post on the Times site at 3:01 a.m., Katches says, “when there is no staff actually working…”

“That type of story is the only exception for when we will publish something immediately. Everything else is read by an editor before posting.”

Anyone (or any bot for that matter) can read a weather report from the NWS, and getting this information out is vital in an emergency situation. But there’s a reason why the public looks to local meteorologists and trusted climate reporters to explain and give nuance to this information, especially during hurricane season.

The Times hasn’t decided if the bot will publish hurricane alerts, Katches said, “But it’s a possibility.”

Denis Philips, the longtime suspender-wearing Chief Meteorologist at Tampa’s WFTS, told CL he doesn’t see the harm in the Tampa Bay Times using an AI bot to scrape the NWS site.

“It’s really no different than the crawls that run at the bottom of the screen of your TV during severe weather,” Philips wrote in a text message. “Now, if these stories go one step further and start addressing POTENTIAL severe weather, that’s another story entirely.”

According to Philips, the NWS and National Hurricane Alerts do in fact offer some detailed information and nuance to storm tracking, but warns that there’s always a “grey area.”

“You’re not just giving a forecast. You’re taking the person by the hand and helping them through the event,” said Philips. “You’re showing empathy and understanding, often because you and your family and friends are ALSO going through it. That sincerity builds trust…something AI simply can’t accomplish.”

The Times is hardly the first publication in Florida to feature AI bylines. In 2021, the Miami Herald launched the Miami Herald Bot, which was also created by United Robots, and has been used in various forms over the years—covering everything from real estate, storms, and even food reporting.

As of now, the paper only acknowledges the use of its AI bot in a disclaimer footer found at the bottom of every Times AI News Engine story. But the Times has yet to create a publicly available AI ethics policy, which is something heavily recommended by the paper’s parent company Poynter.

“That literally is in the final editing stages,” Katches said.

Pitch in to help make the Tampa Bay Journalism Project a success.

This article appears in Feb. 19 – 25, 2026.