Media Trends Executive

topics

politics-policy

epstein-files

donald-trump

2026-midterm-elections

immigration

behind-the-curtain

race-and-equity

technology

privacy-tech-regulation

automation-and-ai

tiktok-news

big-tech

social-media

cybersecurity

economy

social-security

inflation

tariffs

federal-reserve

jobs

transportation

business

stock-market-news

retail-news

crypto

media

elon-musk

dealmaking

health

rfk-jr

drug-prices

axios-explains-abortion

public-health

world

israel

axios-explains-ukraine

iran

russia

china

energy-climate

nuclear-energy

climate-energy-policy

climate-change

extreme-weather

alternative-energy

electric-vehicles

science

emerging-tech

biomedicine

newsletters

charlotte

detroit

columbus

austin

kansas-city

dallas

atlanta

washington-dc

nw-arkansas

boulder

chicago

indianapolis

nashville

richmond

houston

san-francisco

seattle

colorado-springs

boston

des-moines

salt-lake-city

philadelphia

denver

phoenix

new-orleans

miami

twin-cities

huntsville

raleigh

san-antonio

portland

san-diego

tampa-bay

cleveland

pittsburgh

axios-local

axios-pro

axios-live

the-axios-show

Media Trends

Axios Media Trends

A network diagram showing select publisher deals and lawsuits with AI companies. The diagram is very elaborate, showing how complex this space is. Notable portions include 15 lawsuits against Cohere. OpenAI and ProRata AI have made 18 deals apiece with select publishers.

Axios Visuals: 2025-08-29-ai-graph

Data: Axios research; Chart: Erin Davis/Axios Visuals

The incredible volume of both deals and lawsuits between publishers and AI companies over the past two years suggests a fundamental shift from the Web 2.0 era to the Agentic Web.

Why it matters: Intellectual property owners have a stronger copyright case against AI firms than they ever did against social media giants, which has empowered them to wage lawsuits or demand larger payouts from Big Tech than they ever could before.

How it works: In the U.S., IP owners who have filed individual copyright registrations for their works can be legally qualified to receive up to $150,000 in statutory damages per infringed work.

Those damages, depending on the volume of works they can prove were illegally used by AI companies to respond to user queries or train their large language models, could amount to billions of dollars if a court or judge sides in their favor. 

That legal dynamic has provided IP owners like the New York Times, News Corp and others with much stronger financial incentive and confidence to be able to sue AI companies.

Already, some lawsuits have proven successful in forcing AI companies to pay out IP owners based on statutory damages. Anthropic on Friday agreed to pay at least $1.5 billion to a group of authors and publishers in the largest copyright settlement in U.S. history.

State of play: These legal efforts have forced AI firms to consider brokering payout deals to content firms, largely as an effort to skirt potential legal or regulatory scrutiny. 

The size of most deals, therefore, can be seen as a reflection of an IP owner's potential legal case against a certain AI company, rather than the intrinsic value of its content.

That's why some massive content companies, especially those focused on digital content, have not seen deal interest from AI companies.

Prior to a 2024 procedural update at the Copyright Office, digital firms had a much harder time registering individual works for copyright protection. Many have a weaker case for statutory damages compared with decades-old print publications as a result.

Between the lines: Some major lawsuits, such as the Times' case against OpenAI and its investor Microsoft, could take years to resolve. But in the interim, the outcomes from other cases have already started to shape new deals.

Reuters' landmark win against competitor Ross Intelligence earlier this year set a new precedent around whether AI companies can use real-time IP that they don't own to fuel their products, not just to train their LLMs. 

Whereas initial deals between AI companies and content owners centered around using content to train LLMs, the Reuters case pushed more AI companies to consider deals that involve provisions around Retrieval Augmented Generation (RAG), or the process by which LLMs provide answers based on real-time or specific data sets.

Compared to 2024, many deals struck this year focus on RAG, which helps make LLMs more accurate and can reduce hallucinations, or made-up, incorrect answers.

Where it stands: Today, most IP owners understand that striking deals that allow AI firms to use their content to train LLMs has become moot, as most of their content has already been accessed for this purpose. 

Instead, many are choosing to broker RAG deals, or are looking to strike agreements where they are compensated by usage, instead of receiving blanket payments upfront. 

Yes, but: For many publishers, usage deals with AI companies may not be as lucrative in the short term.

While most IP owners are invested in finding long-term solutions for fair compensation, building a two-sided marketplace where that will be possible is extremely challenging because it requires the buy-in of most major AI and search firms.

Some AI companies, like Perplexity, have started to introduce new models that pay out publishers on a usage basis, but those solutions are very hard to scale across the entire web. 

The big picture: A two-sided marketplace can't thrive without participation from the biggest AI and search companies, and right now, there isn't enough legal incentive.

Most notably, Google — the biggest search company in the world by far — faces little legal pressure from IP owners that are still heavily reliant on its search engine for traffic.

This past week, a U.S. federal judge avoided imposing harsh antitrust penalties on Google after it lost a landmark case around its search practices.

The bottom line: AI companies understand the legal challenges ahead if they don't embrace a future where IP owners are fairly compensated for their work, but few feel threatened enough by content companies right now to do all that much about it.

Copyright Office remains in flux after director's firing

Hollywood's AI dilemma calls for courtrooms and collaborations

ChatGPT is still by far the most popular AI chatbot

AI marketplaces multiply to scale publisher deals

Lawmakers draft copyright legislation as AI regulation drifts in limbo

Copyright threats force AI firms to consider stronger deals with publishers

Intellectual property owners have a stronger copyright case against AI firms than they ever did against social media giants.

Artificial Intelligence

Big Tech

Media

The incredible volume of both deals and lawsuits between publishers and <a class="gtmContentClick" data-vars-link-text="AI companies" data-vars-click-url="https://www.axios.com/technology/automation-and-ai" data-vars-content-id="289a8965-7ae7-4675-bdf0-6ca2b92c30f4" data-vars-headline="Copyright threats force AI firms to consider stronger deals with publishers" data-vars-event-category="story" data-vars-sub-category="story" data-vars-item="in_content_link" href="https://www.axios.com/technology/automation-and-ai" target="_self">AI companies</a> over the past two years suggests a fundamental shift from the Web 2.0 era to the Agentic Web.Why it matters: Intellectual property owners have a stronger copyright case against AI firms than they ever did against social media giants, which has empowered them to wage lawsuits or demand larger payouts from Big Tech than they ever could before.

Latest news on Wall Street, Corporate America and financial regulation.

Business

From the Valley to D.C., Big Tech players like Facebook, Google and Amazon are under more scrutiny than ever as new technology develops and privacy and antitrust concerns grow in lockstep with companies’ ambitions.

Copyright threats force AI firms to consider stronger deals with publishers

What to read next