Every morning around 9AM ET, CNET<\/em> publishes<\/a> two stories listing the day\u2019s mortgage rates and refinance rates. The story templates are the same every day. Affiliate links for loans pepper the page. Average rates float up and down day by day, and sentences are rephrased slightly, but the tone \u2014 and content \u2014 of each article is as consistent as clockwork. They are perfectly suited to being generated by AI.<\/p>\n<\/div>\n The byline on the mortgage stories is Justin Jaffe<\/a>, the managing editor of CNET Money<\/em>, but the stories aren\u2019t listed on Jaffe\u2019s actual author page<\/a>. Instead, they appear on a different author page that only contains his mortgage rate stories. His actual author page lists a much wider scope of stories, along with a proper headshot and bio.<\/p>\n<\/div>\n CNET<\/em> is the subject of a swirling controversy around the use of AI in publishing, and it\u2019s\u00a0Jaffe\u2019s team that\u2019s been at the center of it all. Last week, Futurism <\/em>reported<\/a> that the website had been quietly publishing articles written using artificial intelligence tools. Over 70 articles<\/a> have appeared with the byline \u201cCNET Money Staff\u201d since November, but an editorial note about a robot generating those stories was only visible if readers did a little clicking around.\u00a0<\/p>\n<\/div>\n It wasn\u2019t just readers that were confused about what stories on CNET<\/em> involve the use of AI. Beyond the small CNET Money<\/em> team, few people at the outlet know specific details about the AI tools \u2014 or the human workflow around them \u2014 that outraged readers last week, according to current and former staffers who spoke to The Verge <\/em>on the condition that they remain anonymous. Under the two-year-old management of a private equity company called Red Ventures, CNET<\/em>\u2019s editorial staff has often been left wondering: was this story written by AI or a co-worker? Even today, they\u2019re still not sure.<\/p>\n<\/div>\n Daily mortgage rate stories might seem out of place on CNET<\/em>, slotted between MacBook reviews and tech news. But for CNET<\/em> parent company Red Ventures, this SEO-friendly content is the point.<\/p>\n<\/div>\n CNET<\/em> was once a high-flying powerhouse of tech reporting that commanded a $1.8 billion purchase price when it was acquired by CBS in 2008. Since then, it has fallen victim to the same disruptions and business model shifts as the rest of the media industry, resulting in CBS flipping the property to Red Ventures for just $500 million in 2020.<\/p>\n<\/div>\n This type of SEO farming can be massively lucrative<\/p>\n<\/div>\n<\/div>\n Red Ventures\u2019 business model is straightforward and explicit: it publishes content designed to rank highly in Google search for \u201chigh-intent\u201d queries and then monetizes that traffic with lucrative affiliate links. Specifically, Red Ventures has found a major niche in credit cards and other finance products. In addition to CNET<\/em>, Red Ventures owns The Points Guy, Bankrate<\/a>, and<\/a> CreditCards.com<\/a>, all of which monetize through credit card affiliate fees. The CNET<\/em> AI stories at the center of the controversy are straightforward examples of this strategy: \u201cCan You Buy a Gift Card With a Credit Card?\u201d and \u201cWhat Is Zelle and How Does It Work?\u201d are obviously designed to rank highly in searches for those topics. Like CNET<\/em>, Bankrate and CreditCards.com have also published AI-written articles about credit cards with ads for opening cards nestled within. Both Bankrate and CreditCards.com directed questions about the use of AI to Lance Davis, the vice president of content at Red Ventures; CNET<\/em>\u2019s disclosure also included Davis as a point of contact until last week.<\/p>\n<\/div>\n This type of SEO farming can be massively lucrative. Digital marketers have built an entire<\/a> industry<\/a> on top of credit card affiliate links, from which they then earn a generous profit. Various affiliate industry sites estimate the bounty for a credit card signup to be around $250 each. A 2021 New York Times<\/em> story<\/a> on Red Ventures pegged it even higher, at up to $900 per card.<\/p>\n<\/div>\n Viewed cynically, it makes perfect sense for Red Ventures to deploy AI: it is flooding the Google search algorithm with content, attempting to rank highly for various valuable searches, and then collecting fees when visitors click through to a credit card or mortgage application. AI lowers the cost of content creation, increasing the profit for each click. There is not a private equity company in the world that can resist this temptation.<\/p>\n<\/div>\n The problem is that there\u2019s no real reason to fund actual tech news once you\u2019ve started down that path.<\/p>\n<\/div>\n On CNET<\/em> senior editor Rae Hodge\u2019s last day, she sent a goodbye email to hundreds of her co-workers imploring them to look more skeptically at their AI co-workers. Her email began with a screenshot of a ChatGPT-generated resignation letter. \u201cI am writing this letter using AI-generated content,\u201d the note reads. \u201cWhile I may not have personally composed these words, I hope they convey the sincere appreciation I have for my colleagues and the work we have done together.\u201d<\/p>\n<\/div>\n In the email, obtained by The Verge<\/em>, Hodge goes on to direct colleagues to ask pointed questions of several <\/em>Red Ventures executives, saying that unattributed AI-written content was being sent to subscribers of a cybersecurity email newsletter. What\u2019s worse, the newsletters had errors in them that \u201ccould cause direct harm to readers,\u201d Hodge wrote in the email.<\/em>\u00a0<\/p>\n<\/div>\n A former CNET<\/em> employee says that Red Ventures was using automated technology for content long before the AI byline began cropping up in November. They say a tool called Wordsmith \u2014 nicknamed \u201cMortgotron\u201d internally because of its use in mortgage stories \u2014 has been used for at least a year and a half.<\/p>\n<\/div>\n \u201cSometimes the Money writers write like they\u2019re bots, too.\u201d<\/p>\n<\/div>\n<\/div>\n But the siloed nature of the teams across CNET<\/em> and Red Ventures has made it difficult for journalists at the site to understand the chain of command \u2014 in particular, who\u2019s using what tools and when. Those who knew of the AI tool and its uses say that the workflow was so unclear, they sometimes couldn\u2019t distinguish between AI-written stories and articles written by colleagues.<\/p>\n<\/div>\n \u201cIt\u2019s used most often in relation to relaying updated mortgage and refinance rates,\u201d said one source who was familiar with the tool. \u201cI was told that it was always effectively a bot writing these stories.\u201d<\/p>\n<\/div>\n Are you a former or current CNET\/Red Ventures employee? <\/strong>I\u2019d love to hear from you. Contact me at mia@theverge.com and I\u2019ll share my Signal. <\/p>\n<\/div>\n<\/div>\n<\/div>\n But even though some on staff knew automation tools were part of the workflow, the scope of their use was unclear to colleagues whose bylines were appearing on the same site. The former staffer says that by the time stories were published on the site, they didn\u2019t always know if AI tools were involved in the production.<\/p>\n<\/div>\n \u201cSometimes the Money writers write like they\u2019re bots, too, and they\u2019re regular humans,\u201d a former employee says. \u201cThe quality of writing is nearly indistinguishable. That does not make it good.\u201d<\/p>\n<\/div>\n But the robot articles published on CNET<\/em> don\u2019t need to be \u201cgood\u201d \u2014 they need to rank highly in Google searches so lots of people open them and click the lucrative affiliate marketing links they contain.<\/p>\n<\/div>\n It was a way to generate content that would take human writers longer \u2014 the \u201cdull SEO-friendly topics\u201d<\/p>\n<\/div>\n<\/div>\n CNET<\/em> staff was notified last fall that some articles would be written by AI, but by the time they found out, several stories had already been published on other Red Ventures websites, according to one staffer. Those sites also lack clarity around what exactly AI is being used for. On Bankrate, one article<\/a> originally published in May was bylined by a human writer<\/a>; it\u2019s now been updated to list an AI author. The content of the story, though, is the same.<\/p>\n<\/div>\n \u201cI don\u2019t know that it was announced in any kind of grand way,\u201d a CNET<\/em> staffer told The Verge<\/em>. \u201cIt just sort of showed up.\u201d<\/p>\n<\/div>\n The justification for the tool given to staff, multiple people say, was that it was a way to generate content that would take human writers longer \u2014 handling the \u201cdull SEO-friendly topics\u201d or making sure that legal requirements for writing about finance are met. It was sold as a way to free up staff time so they could do more thoughtful work. Instead, several staffers have departed since November, and morale is low at the outlet after several rounds of layoffs, according to former employees.<\/p>\n<\/div>\n Reached for comment, Red Ventures refused to answer any questions about the AI tools it uses, the types of content it generates, or how it disclosed the practices to readers. Instead, an unnamed spokesperson directed The Verge<\/em> to CNET<\/em> editor-in-chief Connie Guglielmo\u2019s note<\/a> defending the use of AI tools at the outlet.<\/p>\n<\/div>\n Using AI in journalism has a much longer history than CNET<\/em>\u2019s ventures. The Associated Press<\/em> was one early adopter and announced it would start using Wordsmith in 2014<\/a> to produce short articles about companies\u2019 earnings reports. In 2016, it expanded coverage to include sports reporting<\/a>, and it now partners with another AI writing firm called Data Skrive for this content<\/a>.<\/p>\n<\/div>\n As with CNET<\/em>, The Associated Press<\/em> frames its use of AI as a way to \u201cfree journalists to do more journalism and less data processing.\u201d The stories it\u2019s automated are high-volume and formulaic, work that bores journalists but is a necessary backbone for wire services like the AP<\/em>. The AP<\/em> labels some stories as automated<\/a> with a footer noting the use of automated tools to create the story, but a reader may not understand what it means that the story was created \u201cusing technology.\u201d<\/p>\n<\/div>\n Red Ventures\u2019 experiments with AI content reflect improvements in the world of AI since 2014. A new breed of AI language models<\/a> is able to easily generate text that is more coherent and covers a wide array of subjects. Studies have found<\/a> that humans are unable to consistently distinguish between text written by humans and the latest AI systems, leading to exactly the sort of confusion CNET<\/em> sources have described. Although Red Ventures has refused to offer more details about the tools it\u2019s using, job descriptions of the company\u2019s staff suggest it is indeed tapping the latest generation of technology and applying it widely.\u00a0<\/p>\n<\/div>\n The company\u2019s \u201c\u200b\u200bcontent head for all AI and automated solutions\u201d for educational websites, Kevin Hughes, says on LinkedIn that he uses not only Wordsmith but also OpenAI\u2019s GPT series \u201cto generate programmatic SEO and bespoke AI content across a dozen websites, generating millions of dollars in revenue.\u201d Hughes lists a number of Red Ventures websites that he\u2019s worked on, including bestcolleges.com, nursejournal.org, and cyberdegrees.org.\u00a0<\/p>\n<\/div>\n With improvements in AI language models over the past few years, experts have warned about potential malicious use cases. Some of the more exotic include automated propaganda and influence peddling<\/a>, but the more prosaic include mass-produced spam and marketing copy. In 2021, Fabian Langer, the founder of an AI writing startup named AI Writer, told The Verge<\/em><\/a> how his tools were already being used to fill \u201cSEO farms\u201d with content. Said Langer: \u201cFor these [SEO] farms, I do not expect that people really read it. As soon as you get the click, you can show your advertisement, and that\u2019s good enough.\u201d<\/p>\n<\/div>\n The cheapness and ease with which these tools can generate content has led some to predict<\/a> that this writing could slowly take over the web, polluting search results and social media with text designed only to push someone to a specific product or website.<\/p>\n<\/div>\n