The first 100 days of Trump to fight against the press, the passage from the media to podcast videos and even more.

The evolution of the AI ​​content license transactions

The Washington Post has become the last major publisher to conclude a license agreement with OPENAI, joining a growing cohort which now covers more than 20 press organizations.

This is part of a familiar model: every few months, Openai locks in another publisher to strengthen its content pipeline. But the terms of these agreements seem to evolve quietly – subtly moving away from the explicit language around the training data which defined previous transactions and raising new questions about what these partnerships now mean.

The agreement of the Washington Post focuses on the overhaul of its content in response to requests related to the news. “As part of this partnership, Chatgpt will display summaries, quotes and links to the original posts of the post in response to the relevant questions,” reads the announcement on April 22 on the agreement of the post with Openai. On the other hand, the past deals with publishers like Axel Springer and Le Temps – signed respectively in December 2023 and June 2024 – explicitly Included provisions for the formation of the OPENAI LLM on their content.

The OPENAI Agreement of Guardian, announced in February 2025, has a wording similar to the announcement of the Washington Post and no mention of training data. A Guardian spokesperson refused to comment on the conditions of the agreement with OPENAI. The Washington Post did not respond to requests for comments.

These somewhat subtle changes in the language language could point out a broader change in the AI ​​landscape, according to conversations with four Legal media experts. This could point out a change in the way in which AI content license transactions are structured in the future, with more publishers who potentially seek agreements that prioritize the allocation and prominence in AI search engines on model training rights.

Another factor to consider: these AI companies have already formed their LLMS on large quantities of content available on the web, according to Aaron Rubin, partner of the Strategic Group Transactions & Licensing of the Gunderson Dettmer law firm. And because IA companies are faced with the disputes of media companies saying that it was a copyright violation – such as the New York Times affair against Openai – if AI companies should continue to pay to obtain data on data for training purposes, this could be considered “implicit admission” which should have paid to dismiss these data, and not to scratch them free. said Rubin.

“”[AI companies] already have a word of words they stolen. They don’t need additional words that for training, but they want to have the updated content for the answers [in their AI search engines]”Said Bill Gross, founder of the AI ​​startup, Prorata.ai, who builds technological solutions to compensate for publishers for the content used by AI generating companies.

According to Rubin, AI companies and publishers can benefit from this potential development. AI companies have access to reliable and up -to -date news from sources of trust to answer questions about the current events of their products, and publishers “can potentially fill a gap in which they were afraid would be missing with the way these AI tools have evolved. Having better attribution in places like Chatgpt research has the potential to generate more traffic to publishers’ sites. At least it is hope.

“He has the potential to generate more money for publishers,” said Rubin. “Publishers put a bet that this is how people will interact with the media in the future.”

Since last fall, Openai has challenged research giants like Google with its AI search engine, Chatgpt search – and this effort depends on access to news content. When asked if the structure of the Openai agreements with the publishers had changed, an Openai spokesperson underlined the launch by the Chatgpt research company in October 2024, as well as the improvements announced this week.

“We have a direct flow to the content of our publisher partner to display summaries, quotes and links attributed to the original reports in response to the relevant questions,” said the spokesperson. “It is a component of transactions. Post-training training helps increase the accuracy of the contents of a publisher. ” The spokesperson did not respond to other requests for comments.

The quantity of publishers such as the Washington Post Stand to be concluded from offers with OpenAi is not clear, especially since a different model can emerge on chatgpt research. But the prospects for license agreements between publishers and IA companies seem to be aggravated. The value of these transactions is “falling”, at least according to the CEO of the Atlantic, Nicholas Thompson, who spoke during the Reuters Next event last December.

“There is always a market for content licenses for training and it is always important, but we will continue to see the focus on the conclusion of transactions which lead to the conduct of traffic to sites,” said John Monteubio, partner of the media group and advanced technologies of the law firm Loeb & Loeb. “It will be the new form of reference marketing [generative AI] tools.”

What we heard

“We don’t have to worry about a somewhat false story of: cookies must go…. Then you can really put this whole bandwidth and power to improve the current market, without worrying about a future potential problem that was in Google’s control from the start.”

– Anonymous publication executive on Google’s decision last week to continue using third -party cookies in Chrome.

Numbers to know

$ 50 million: the amount that Los Angeles Times lost in 2024.

50%: The percentage of American adults who have said that AI will have a very or somewhat negative impact on the news that people get to the United States in the next 20 years, according to a study by the Pew Research Center.

100 million dollars: The Spotify amount has paid publishers and creators of Podcast since January.

0.3%: the expected drop in the use of media (digital and traditional channels) in 2025, the first drop since 2009, according to PQ Media Research.

What we have covered

AI proceedings highlight the difficulties of publishers to block the boots in the scratch of content

• The recent trial of Ziff Davis against Openai highlights the reality that publishers still have no reliable means of preventing AI companies from scratching their content for free.
• While tools such as robots.txt files, paywalls and blocking tags have emerged, many publishers admit that it is very difficult to apply control over each bot, especially since some people ignore the standard protocols or mask their identity.

Learn more here.

Who would buy Chrome?

• The Google Search Antitrust trial could force Google to separate from the Chrome browser.
• If that was the case, Openai, Perplexity, Yahoo and DuckduckGo could be some of the potential buyers.

Find out more about the potential impact of a chrome sale here.

Tiktok courts creators and agencies to participate in his live tools

• Tiktok tries to prove the income potential of his live tools.
• The social media platform claims that its creators now collectively generate $ 10 million in revenues daily thanks to live broadcast.

Learn more in Tiktok’s field here.

WTF are gray bots?

• Some are called generative robots from AI and gray scrapers to illustrate the blurred line between real and false traffic.
• These robots can have an impact on analysis and steal content, and IA -focused advertising prints can harm click and conversion rates.

Find out more about the reasons why gray boots are a risk for publishers here.

Does Facebook become a new source of income for publishers again?

• The publishers attended a recent peak of one year on the other in Facebook reference traffic, and it is – somewhat surprisingly – coinciding with an influx of income from the Meta content monetization program.
• Of the 10 publishers, Digiday spoke for this article, several are on the right track to make between six and seven figures this year of the Meta content monetization program.

Learn more about what publishers get from Facebook here.

What we read

The video ambitions of the News Outlet podcast highlight the passage from audio format to television

The media like the New York Times and the Atlantic put more resources to the production of videos from popular podcast emissions to exploit the youngest YouTube audience, Vanity Fair reported.

Perplexity wants to collect data on users to sell personalized ads

Perplexity CEO, Aravind Srinivas, said Perplexity builds its own browser to collect user data and sell personalized announcements, Techcrunch reported.

President Trump targets the press in the first 100 days

President Trump targets traditional media companies during his first 100 days, using tactics such as the prohibition on outlets to cover the events of the White House to launch investigations on the main networks, Axios reported.

SEMAFOR will test the subscriptions

SEMAFOR “will try” subscriptions in “time”, told the founder Justin Smith to the intelligence of the New York Magazine in a deep dive in the news start-up focused on the newsletter.