EXCLUSIVE: The anchor is in studio, speaking directly to the camera then cuts to a news segment, one in which the words coming from a French-speaking man come out of his mouth as English.
This is one of the elements of an AI-produced newscast being introduced later today by startup Channel 1, set to run on its own site, Channel1.ai, as well as Crackle, Redbox Free Streaming and X/Twitter.
More from Deadline
The Channel 1 startup, from entrepreneur Adam Mosam and producer Scott Zabielski, has a distribution deal with Chicken Soup for the Soul Entertainment, and the plans are to introduce a FAST channel in February or March followed by mobile and connected TV apps, the latter of which will offer personalization of viewers’ news diets.
Mosam, a tech entrepreneur who has been chief digital officer for Chicken Soup, said that, in contrast to previous, Max Headroom-like experiments, the technology “has gotten to a place where it’s comfortable to watch.”
“Not only is the visualization there, someone that can speak to you and deliver the news, but there are so many other base technologies that we can leverage to further enhance the experience of watching the news,” he said.
What’s debuting this evening is what Mosam calls the “showcase episode,” a pilot “to demonstrate everything from editorial to technology. We just thought it was important to prove we could execute on the idea and show this off,” he said. He said that the project so far is self funded, but they will raise capital shortly. “
Channel 1, he said, will draw on content from three separate sources: They’ll be partnering with a yet-to-be-announced news agency, drawing on content from independent journalists, and creating AI-generated news from “a trusted primary source,” like a government document or SEC report, “that we can put together and format for our audience.”
“We are not gathering primary data sources, but we are processing that data,” Mosam said.
The initial newscast will be a mixture of licensed clips and fair use content, while the anchors will actually be based on real people who have been compensated for the use of their likenesses, he said.
The debut of the pilot comes as networks, like CBS News, are devoting more resources to identifying AI deepfakes and other misinformation, to the point of hiring so-called “forensic” journalists.
For its part, Channel 1 plans to feature an icon on the corner of the screen that will identify portions that are generated. For example, the icon pops up when the French man’s words are translated into English.
“I think people are correct in their thinking that we need to be cognizant and aware of this problem,” Mosam said. “This is going to be an issue, as a society, we are going to have to deal with, because as the technology improves, it will get easier to generate what we would maybe term ‘fake’ or ‘fake news’ or generate itself.”
He said that in their content, “they’re very clear that the audience at every point in time knows exactly what they are looking at if there has been some sort of AI modification.” He also said that “at every step of our editorial process, we do have humans in the loop,” with plans to hire an editor in chief within the next six months. A staff of 11 people are working on the project right now.
“We’re not coming at this from, ‘Oh, we’re targeting these major cable news networks,” Mosam said. ‘We’re coming at this from more of a first principles idea, understanding that this technology exists, and as digital natives ourselves, we understand how to build that technology. We understand what to be worried about.” He said that plans also are in the works is a system that will provide users’ with proof of source material, “almost like a chain of custody of where we found our news.”
Zabielski said, “We want the audience to feel comfortable that this news is coming from trusted sources and that we are just formatting this in a way, and distilling it down, in a way that fits our format to present it to the audience. People feeling comfortable that this data is trustworthy is number one, because if people don’t trust it, they they’re not going to want to watch our news.”
That said, the use of AI has caused anxiety throughout the news business, not just over the IP questions raised by the use of content in training models, but for the prospect of vast numbers of employees being displaced. AI was a sticking point in the recently resolved WGA and SAG-AFTRA strikes, which had to do with film and TV contracts, while newsrooms have set up guidelines for the use of the technology for the time being.
Mosam said that the production costs of Channel 1 will be much cheaper than traditional cable news stations. And they don’t doubt that the technology poses a threat to the talking heads of today.
Zabielski said that for the anchors who merely read off Teleprompters, “it’s hard to imagine the technology isn’t going to come along at some point for those people. But the reality is most of the people in the news business are also journalists and reporters that also appear on camera. And that’s not what we are trying to replace. Those are the people who are doing the actual reporting. We still need that reporting to be done, whether or not an avatar is delivering that or we’re the ones going on camera.”
He noted that a news producer “could have a trusted avatar where they become trusted reporter without having to go on camera, but it’s really their work, but they are getting compensated and their work is still being seen.” He also suggested that content licensing could provide a revenue stream for local news outlets, which have been hollowed out across the country.
Mosam also said that they are developing an anchor that would “have significantly more depth” than the AI-generated personalities that already are appearing in Korea, India and China. “So if you can think of this anchor, it is not just, now you fed it a script and its lips moved. It has a significant amount of technology behind it, where if you can think that every anchor would be its own version of Chat GPT. It would have its own memory capability. It would have its own linguistics, its own tone of voice.” It also will be tailored for different parts of the world.”
The ultimate goal of Channel 1 is a personalized app, where users can customize everything from their news interests to the “style of the anchor you like to deliver the news to you,” Zabielski said.
Such personalization of the news has also been the source of some consternation, with audiences drawn to personalities and voices that they want to hear, creating their own information bubble. But that’s a longtime trend, via cable TV and social media, that predates AI.
The app, Mosam said, will allow users to “navigate through the news similarly in a way that someone may be watching TikTok, for example. And because you are making decisions so quickly, as opposed to watching a streaming service, traditionally where you are watching a TV show or a movie, we will learn about you fairly quickly and understand what it is you are looking for and what type of news that we can help to inform you.”
Best of Deadline