Media leaders are sharpening their language as artificial intelligence giants lean heavily on news content, with one of the world’s biggest wire services calling the situation “existential”. At the Andrew Olle Lecture in Sydney, its editor-in-chief warned that AI firms are raiding copyrighted journalism to feed their models.
At the centre of the complaint is how large language models are trained. AI companies ingest vast libraries of articles, photos and videos from news organisations, which are protected by copyright and cost significant resources to produce.
Media executives argue that this content is being used at scale without meaningful consent or payment. AI systems gain commercial value from material that originally relied on newsroom budgets and subscriber revenue.
Industry groups point out that journalism involves reporters, editors, legal checks and on‑the‑ground verification, all funded by advertising and subscriptions that are already under pressure. When AI tools reproduce or repackage this work, users may never visit the original sites, weakening traffic and ad revenue further.
That erosion of audience relationships and income makes it harder to pay for investigations, foreign bureaus and specialist beats. The fear is a vicious cycle where fewer resources lead to thinner coverage, even as AI systems keep training on past work.
The broader anxiety runs beyond copyright law and licensing deals. Media organisations worry that if AI platforms become the primary way people access information, tech companies will capture most of the economic value of journalism while controlling distribution.
Many publishers question whether they can negotiate sustainable terms with AI firms before financial damage to original reporting becomes irreversible.
