What Happened

Australian media companies are facing a stark choice: negotiate commercial licensing agreements with major AI developers, or watch years of journalism get ingested into large language models without compensation.
The warning came into sharp focus after executives from Australia's largest media organisations spoke at the Andrew Olle Lecture, detailing their companies' thinking on income in a new age of reporting. News Corp Australasia executive chairman Michael Miller was among the most direct, describing AI data mining in terms that stopped just short of theft. "I don't know if theft is the right word, but it's certainly the case that they are bypassing the consent of the news organisations in doing this," Miller said, according to IT Brief Australia's reporting of the event.
Miller argued that further revenue decline would mean fewer journalists at The Daily Telegraph, Herald Sun and The Australian, eroding the capacity of those mastheads to hold power to account. "But we do know that even though news organisations have flagged that they're willing to work with these organisations, they bypassed that opportunity and just decided to go for the free route, anyway," he said.
The issue has also reached the highest levels of federal politics. In a recent speech on AI, Prime Minister Anthony Albanese stressed that copyright across the media industry must be respected by companies including OpenAI and Anthropic.
Why It Matters
The scraping question is not abstract. AI companies train large language models on vast text corpora, and decades of Australian journalism represent exactly the kind of high-quality, factual writing those models prize. If that content is ingested without a licensing arrangement, news organisations receive nothing while AI products built on their work generate commercial returns.
Professor Terry Flew, Professor of Digital Communications and Culture at Sydney University, framed the broader tension plainly. "There are some fascinating tensions in how news and journalism deals with AI," he said. He also drew a line between what AI can replicate and what it cannot: "There is a difference between reporting the result and being able to comment on the game, the teams and the players."
That distinction matters commercially. Raw facts about a sporting result are not protected, but the analysis, context and voice that surround them are. Flew noted that the question of audience demand for media "always has to be tied with the question of preparedness to pay," pointing to the unresolved challenge of monetising journalism in an era when AI can summarise it for free.
Key Details
IT Brief Australia reported that the industry is already splitting into two camps. Many publishers have accepted that licensing agreements with AI developers are becoming an inevitable part of the industry's future. Others, particularly some legacy mastheads, have refused that path entirely.
The deals being struck vary in structure and value, and the terms are rarely disclosed publicly. What is clear, according to the reporting, is that the window for negotiating from a position of strength may be closing. "We're moving into a stage where deals are being made with AI companies around rights to ingest content into large language models," one observation from the lecture noted. "But a lot of news organisations are cutting deals with AI companies."
Background and Context
The tension between AI developers and news publishers is not unique to Australia, but the country's concentrated media ownership and relatively small advertising market make the stakes acute. Australian Consumer Law provides some baseline protections around misleading conduct, but there is no specific federal framework governing AI training data and copyright. The Albanese government's intervention signals that regulatory attention is growing, though no legislation has been announced.
The Andrew Olle Lecture, named after the late ABC broadcaster, has long served as a forum for serious reflection on the state of Australian journalism. Its use as a venue for this debate reflects how central the AI licensing question has become to the industry's self-understanding.
What Comes Next
The federal government's position is clear in principle: AI companies must respect media copyright. Translating that principle into enforceable rules is the harder task. Industry bodies and individual publishers are likely to keep pressing for a formal framework, while AI developers will continue signing deals where they judge the cost worthwhile and scraping where they do not.
For news organisations still holding out, the calculus is uncomfortable. Refusing to licence content preserves a point of principle but offers no guarantee of protection. Signing a deal means accepting terms set largely by the other side. Neither path is clean, and the organisations that delay the decision may find it made for them.