AI companies 'stole' Australian culture, news bosses say as they call for licensing deals

The heads of Australia’s biggest media companies have demanded that AI companies pay for the “public good” of Australian news and culture used without permission to train their technology.

Australian news agencies are already in talks for licensing deals, with the ABC’s chief revealing that AI developer Anthropic had approached it about audio and video that capture how Australians speak.

The comments came during a panel featuring the heads of the ABC, SBS, Nine and News Corp Australasia at the annual Andrew Olle lecture on Tuesday night.

The wide-ranging discussion covered everything from the role of AI in journalism to new detail about the departure of former Today show host Karl Stefanovic.

ABC managing director Hugh Marks told the audience that journalism was a public good that needed to be protected.

Mr Marks argued that AI companies should strike deals with Australian news companies to license works, but also acknowledged a “delicate balance” with also providing high-quality information to Australians who are using AI.

This obligation to inform shouldn’t mean that news organisations should be expected to let AI companies train on news free of charge, SBS managing director Jane Palfreyman said: “Our collective view on that would be no.”

News Corp Australasia executive chairman Michael Miller said that AI companies knew that the “theft” of Australian works was a “risk to their business in Australia”.

Nine chief executive and managing director Matt Stanton said his company’s recent licensing deal, under which its journalism will be cited in answers from Microsoft’s AI tool Copilot, was a “good deal for both sides”.

Mr Stanton said that Nine’s decision to cut 30 job was not “directly” as a result of AI, when pressed by panel moderator ABC Radio Sydney Mornings host Hamish Macdonald.

He also said Karl Stefanovic’s “different directions” of his podcast had prompted Nine’s decision to part ways with the former Today host, even before his controversial podcast with Tommy Robinson.

Paying for journalism

The panel discussion followed a lecture by global news agency Reuters’ editor-in-chief Alessandra Galloni, who urged news organisations to set firm terms for artificial intelligence companies using their work and to share what they learn when AI tools go wrong.

Ms Galloni said journalism remains the source of reliable, first-hand reporting on which people and increasingly AI systems depend.

She cited recent research showing that 10 per cent of people access chatbots for news each week, including 16 per cent of people under 35.

Ms Galloni argued publishers had no reason to give up on reaching those audience but should strike deals to have their journalism correctly attributed, accurately represented and paid for by the AI companies.

She cited a speech delivered by Prime Minister Anthony Albanese last week where he promised that Australian writers, musicians, artists and journalists would retain control over work used to build and train AI.

Using AI in the newsroom

Ms Galloni argued that it is the responsibility of news organisations to help society understand the “pitfalls and possibilities of AI.”

She said that Reuters using AI to help reporters search large collections of documents, find patterns and turn official material into fast news updates, summaries and translations.

She cited investigations that used AI to document thousands of social media posts and speeches by US President Donald Trump, and to convert 20,000 pages of Syrian government documents into searchable text that was used in an investigation that uncovered a mass grave.

Reuters is also using AI to produce headlines from press releases, draft first paragraphs in urgent stories, generate bullet-point summaries and translate stories, Ms Galloni said.

But Reuters only published journalism produced with AI once it had been reviewed by a journalist and its use was disclosed when it played a significant role in what was published, she said.

She said on top of the AI’s struggles with providing nuance and context, it is also vulnerable to hallucinations and other mistakes. One third-party tool tested by Reuters had once mistaken a SpaceX spacecraft for a missile, she said.

“Thankfully we caught it,” she said.

Ms Galloni said even advances in AI could not replace the reporting that made news valuable in the first place.

Reuters has 2,600 journalists in 165 countries, she said, including 70 staff covering the war in Ukraine and Russia.

“In my conversations with tech companies, I have often been struck by how little they understand about what original, first-hand, on-the-ground journalism is and looks like,” she said

Sharing what works

Ms Galloni called on rival news organisations to share the lessons emerging from those experiments, drawing a comparison to the way that competing journalists will exchange safety information when in war zones.

“I would venture that it is our collective responsibility — to the public, to society — to enhance our and their understanding about the pitfalls and possibilities of AI,” she said.

The Andrew Olle media lecture was first held in 1996 to honour the ABC radio and television broadcaster who died from an undiagnosed brain tumour in 1995.

Each year’s lecture features a speech on the future of journalism in Australia and raises funds for charity Brain Cancer Australia.

Previous lecturers have included New York Times executive editor Joseph Kahn, Ray Martin and former News Corp co-chairman Lachlan Murdoch as well as ABC journalists Leigh Sales and Fran Kelly.