The Digiday+ Media Briefing is sent out via email every Thursday morning at 10 am ET. ET. More from the series:
The Washington Post’s agreement with OpenAI, to use its content in ChatGPT’s search product, is examined this week. What does it say about the future of licensing agreements between AI companies and publishers?
- The focus of deals between AI companies, publishers and search engines is shifting subtly.
- Trump’s first 100-day battle with the press, the shift of news outlets to podcast videos, and more.
The evolution in AI content licensing
The Washington Post is the latest major publisher that has struck a licensing agreement with OpenAI. The network spans over twenty news organizations.
This is a common pattern: OpenAI signs up a new publisher every few months to boost its content pipeline. The terms of these agreements are quietly changing — they’re moving away from the explicit language about training data that defined previous deals and raising new questions about what these partnership mean now.
The Washington Post’s agreement focuses surfacing its content as a response to news-related questions. As part of this partnership ChatGPT will display summaries and quotes from The Post as well as links to the original reporting in response to relevant queries. The announcement made on April 22 about OpenAI’s deal with the Post reads as follows. Previous deals with publishers such as Axel Springer, Time and Time – which were signed in December 2023 respectively and June 2024 – explicitly provided for OpenAI’s LLM to be trained on their content.
The Guardian announced its OpenAI deal in February 2025. The Washington Post’s announcement uses similar language, but does not mention training data. A Guardian spokesperson declined comment on the terms of the deal with OpenAI. The Washington Post did respond to requests for comments. According to four legal experts in media, these subtle changes in the language could indicate a wider change in the AI landscape. It could mean a shift in how AI content licensing agreements are structured in future. More publishers may seek agreements that prioritize attribution, prominence in AI search engine and rights for model training over rights to model training. Aaron Rubin, partner at Gunderson Dettmer’s Strategic Transactions & Licensing Group, said that these AI companies had already trained their LLMs using vast amounts of web content. Because media companies have sued AI companies for copyright infringement, such as The New York Times case against OpenAI, if AI companies continue to pay for data licensing for training purposes, this could be viewed as “an implied admission” that they should have licensed that data from the beginning, Rubin said.
[AI companies] have already stolen a trillion words. They don’t really need the extra words for training but they do want the updated content [in their AI search engines],” said Bill Gross. He is the founder of AI startup ProRata.ai which builds tech solutions to compensate publishers who use content from generative AI companies. Rubin says that both AI companies and publishers will benefit from this possible evolution. AI companies can access reliable, current news from trusted sources in order to answer questions regarding current events within their products. Publishers, meanwhile, “can potentially fill a hole that they were afraid would be missed with the way these AI tools are evolving.” He said that they were losing clicks, eyeballs, and links to their pages. Better attribution in places such as ChatGPT Search could drive more traffic to publisher’s sites. At least, this is the hope. Rubin stated that the technology has the potential to increase revenue for publishers. Publishers are betting that this will be the way people interact with media in the future.
OpenAI has been challenging search giants such as Google with Openai’s Ai Search Engine, ChatgpT Search ( ), depends on the availability of news content to power its AI search engine. OpenAI’s spokesperson was asked if OpenAI’s contracts with publishers had changed. She cited the launch of ChatGPT in October 2024 as well as improvements. This weekwas announced.
We have a direct feed from our publisher partners’ content to display summaries and quotes, as well as links to original reporting when responding to relevant questions. This is part of the deal. The spokesperson did no respond to any further questions.
It is unclear how much money publishers like The Washington Post will make from OpenAI deals, especially since a new model may focus on ChatGPT search. The outlook for licensing agreements between publishers and AI firms appears to be worsening. The value of these agreements is “plummeting,” according to The Atlantic CEO Nicholas Thompson. Reuters Next event in December.
There is still a market for content licensing for training, and that is still important. But we will continue to focus on entering deals that result in driving visitors to sites,” said John Monterubio. Partner in the Advanced Media & Technology Group at law firm Loeb & Loeb. “It will be the new form for SEO marketing and ad-buying — to appear higher on results when communicating with [generative AI] Tools.”
What we heard
We don’t need to worry about the false narrative that cookies must go …. Then, you can use all of that bandwidth and horsepower to improve the current market instead of worrying about a future problem which was under Google’s control from the beginning.
Numbers to Know
The Los Angeles Times will lose $50 million in 2024.
Spotify has paid podcast creators and publishers $100 million since January.
According to PQ Media Research, 0.3%is the expected decline in media consumption (both digital channels and traditional channels) by 2025. This is the first decline since 2009.
What we’ve covered:
AI suits highlight publishers’ struggles to stop bots from scraping the content
- Ziff-Davis’ recent lawsuit against OpenAI highlights that publishers still do not have a reliable way to prevent AI companies from scraping the content for free. Although tools such as robots.txt, paywalls, and AI-blocking tag have been developed, many publishers admit that it is difficult to enforce control over every bot. This is especially true when some bots ignore standard protocols or hide their identities.
Read on to learn more.
What would make someone buy Chrome? The Google search antitrust case could force Google to give up the Chrome browser. If this happened, OpenAI, Perplexity and Yahoo could be potential buyers.
You can read more about the possible impact of a Chrome sale here.
TikTok wants creators and agencies involved in its livestreaming tool
- TikTok tries to prove that its livestreaming tool can generate revenue.
- According to the social media platform, its creators now generate $10 million in revenue per day through livestreaming.
Read about TikTok’s pitch here.
What are gray bots exactly? Some people call the AI crawlers, scrapers and generators “gray bots”to show how blurred the line is between real and fake traffic. These bots can affect analytics and steal content. AI-driven ads can also harm clickthrough rates and conversion rates. Read more about gray bots and why they are a threat to publishers
.
Could Facebook be a new source of revenue for publishers?
- Publishers are experiencing a spike in Facebook referral traffic over the past year. This coincides, somewhat surprisingly, with an increase in revenue from Meta’s new content monetization program.
- Several of the 10 publishers Digiday interviewed for this article are on track to earn between six and seven figure incomes this year through Meta’s newest content monetization program.
Find out what publishers get from Facebook by clicking here.
What we’re Reading
News organizations’ podcast video ambitions show the audio format’s move to TV
According to Vanity Fair, news outlets such as The New York Times and The Atlantic have increased their resources in order produce video content from popular podcasts. This is done in an effort for them reach younger YouTube audiences.
Perplexity wants data on users for personalized ads
Perplexity’s CEO Aravind Srinivas told TechCrunch that Perplexity was building its own browser in order to collect user data and to sell personalized advertisements.
(19459030) President Trump targets the media in his first hundred days
In his first 100-day tenure, President Trump has targeted traditional media companies, using tactics like banning outlets from covering White House Events and launching investigations against major networks, Axios reports.
Semafor to test subscriptions
Justin Smith, founder of the newsletter-focused start-up news company Semafor, told the New York Magazine Intelligencer that the company will “try out”subscriptions “in due time”.
