Journalist Javier Cabral wanted to test Google’s much-touted experimental artificial intelligence-powered search results, so he typed in a question about a topic he knows a lot about: coffee from Long Beach bakery Gusto Bread.
In less than a second, Google’s AI summarized the bakery in a few sentences and bullet points. But Cabral said the summary was not original and appeared to have been taken from an article he wrote last year for LA Taco, a local food, community and culture publication where he is editor-in-chief. For the previous article, he spent at least five days writing a profile on the bakery, coming in at 4 a.m. to report on the bread-making process.
The way Cabral sees it, the search giant’s AI was deceiving him.
“The average consumer who just wants to go and look is probably not going to read the book. [the article] “I don’t do that anymore,” Cabral said in an interview. “When I analyze it that way, it’s definitely a little infuriating.”
The rise of AI is just the latest existential threat to news organizations like Cabral’s, which are fighting to survive in a rapidly changing media and information environment.
1
2
1. LA Taco editor Javier Cabral in the alley behind Figueroa Theatre in Los Angeles in 2020. (Mariah Tauger/Los Angeles Times) 2. June 26th at the LA Taco offices in Los Angeles. (Zoe Cranfill/Los Angeles Times)
In the internet age, news media have struggled to attract subscribers and advertising dollars, and social media platforms like Facebook, which publishers relied on to reach huge audiences with their content, have pivoted away from news. Now, the growth of AI from companies like Google, Microsoft, and ChatGPT-maker OpenAI has publishers fearing that digital programs could automatically scrape information from archives and distribute it for free to readers, with devastating consequences.
“This is fundamentally unfair,” said Daniel Coffey, president and CEO of the News Media Alliance, which represents publications including The New York Times and the Los Angeles Times. “In a scenario where they were using our stories to compete with us, our business model would become nonexistent. We’re very concerned about that.”
Tech companies leading the AI push say their tools do not contribute to piracy and can drive more traffic to publishers.
Google said in a statement that AI Summary, the summary that appears when a user types a search query, is designed to “provide a snapshot of relevant information from multiple web pages.” The companies also provide links to the summary so users can learn more.
AI and machine learning can be useful tools for publishers to conduct research and make recommendations to readers, but for many journalism media, the AI revolution represents another outcome: tech giants becoming middlemen between content creators and consumers and monopolizing the profits.
“Over the last 20 years, big tech companies have dictated the business model for news and essentially dictated how news is distributed through search and social media, with extremely disastrous consequences for most news organizations,” said Gabriel Kahn, a professor at the University of Southern California’s Annenberg School for Communications and Journalism.
LA Taco operates on a tight budget, and its publisher doesn’t take a salary. The site gets most of its revenue from membership, so if people don’t pay to read LA Taco articles and instead get their information directly from Google, that’s a big problem. Pictured above is a staff meeting in the Chinatown office.
(Zoe Cranfill/Los Angeles Times)
To address the problem, news organizations have taken widely different approaches. Some, including The Associated Press, the Financial Times, and News Corp., which owns The Wall Street Journal and Dow Jones, have signed licensing deals with San Francisco-based Open AI to use their content in exchange for compensation. Vox Media and The Atlantic have also struck deals with the company.
Some have taken the fight to court.
The New York Times filed a lawsuit against OpenAI and Microsoft in December, alleging that the companies used its articles to train their digital assistants and provided users with the text of paid articles for free, actions that the paper estimates have cost it billions of dollars in damages.
Separately, Forbes last month threatened legal action against AI startup Perplexity over alleged plagiarism. After receiving Forbes’ letter, Perplexity said it had changed how it presented sources and adjusted the prompts in its AI model.
The company said it is developing a revenue-sharing program with publishers.
In its lawsuit, the New York Times said it’s fighting AI not just to get paid for content, but also to protect the future of the journalism profession.
“With reduced revenue, news organizations will have fewer journalists to devote time and resources to important, in-depth stories, and as a result, they risk those stories going unreported,” the paper said in the lawsuit. “The output of journalism will be diminished, and the costs to society will be enormous.”
OpenAI said the New York Times lawsuit is without merit and that it could not replicate the instances of ChatGPT repeating paid articles cited by the paper. The company said publishers have a way to opt out of having their sites used to train AI tools. Microsoft did not respond to a request for comment.
The Associated Press, the Financial Times and News Corp, which owns The Wall Street Journal and Dow Jones, have signed licensing agreements that allow San Francisco-based OpenAI to use their content in exchange for compensation.
(Michael Dwyer/The Associated Press)
“Microsoft and OpenAI are completely turning the process backwards,” Davida Brook, a partner at law firm Susman Godfrey, which represents The New York Times, said in a statement. “Neither The New York Times nor any other creator should have to resist having their work stolen.”
The legal battle is widening: In April, eight publications owned by private equity firm Alden Global Capital also accused OpenAI and Microsoft of using and providing information in their news articles for free.
In some cases, OpenAI’s chat tool provided false information that was attributed to publications, Media News Group and Tribune Publishing executive editor Frank Pine said in a statement. For example, Pine said OpenAI said the Mercury News recommended injecting disinfectant to treat COVID-19 and that the Denver Post published a study suggesting smoking helps with asthma. Neither publication made such claims.
“[W]”Instead of reporting actual, verbatim reports by hard-working journalists, they falsely attribute fake information to our news publications and undermine our credibility,” Pyne said.
OpenAI said it was “not previously aware” of Alden’s concerns, and added that it is “actively engaged in constructive partnerships and dialogue with many news organizations around the world to explore opportunities, discuss concerns, and offer solutions.”
One such collaboration is OpenAI’s recent deal with News Corp., which allows the company’s tools to surface content from news publishers in response to user questions and access content from The Wall Street Journal, The New York Post and British and Australian publications to train its AI models. The deal is worth more than $250 million over five years, The Wall Street Journal reported, citing anonymous sources. News Corp and OpenAI declined to comment on financial terms.
“This landmark agreement is not the end, but the beginning of a beautiful friendship that will see us working together to produce and deliver insight and integrity in real time,” News Corp Chief Executive Robert Thomson said in a statement.
“By making it easier for people to discover content through our tools, we strive to foster an ecosystem of publishers and creators,” OpenAI said in a statement.
OpenAI has deals with several publishers, but the tech industry argues that under “fair use” principles, which allow for limited reproduction of content without the copyright holder’s permission, it should be able to train its AI models on online content and surface relevant information.
“As long as these companies aren’t reproducing verbatim what news sites are publishing, we believe it is within their legal right to make this content available to users,” said Chris MacKenzie, a spokesman for the Chamber of Progress, a trade group that represents companies including Google and Meta. “At the end of the day, it’s important to remember that no one owns a copyright on facts.”
But outlets like The New York Times deny these fair use claims, arguing that in some cases the chatbots are replicating their content and unfairly benefiting from their heavily vetted and fact-checked work. For smaller outlets like LA Taco, which can’t afford to sue OpenAI or develop their own AI platform, the situation is even tougher.
Based in LA’s Chinatown, LA Taco has four full-time and two part-time employees, operates on a tight budget, and the publisher doesn’t take a salary. The majority of the site’s revenue comes from memberships, so it’s a big problem if people get their information directly from Google rather than paying to read LA Taco articles.
Another way to address the disruption of journalism by big tech companies is through legislation. The California Newspaper Publishers Association, which includes the Los Angeles Times, has introduced a state bill called the California Journalism Protection Act. The bill would require digital advertising giants to pay news organizations for access to their stories, either a pre-determined fee or an amount determined through arbitration. Most publishers would have to use 70% of the funds they receive to pay reporters’ salaries. Another bill lawmakers are considering would inject funds into news organizations by taxing the data big tech platforms collect from users and giving tax credits to companies that hire full-time reporters.
“We need some kind of regulation to get out of this,” said USC’s Khan. “Congress can’t do anything, so these platforms are basically free to do what they want with almost no consequences.”
Times Editorial Librarian Carrie Schneider contributed to this report.
