Big Tech’s National Security Case for AI Copyright Changes Challenged in New Australian Report

A new report from an Australian defence analyst has pushed back hard against one of the strongest arguments used by large AI companies in Canberra. The report says big tech firms have not shown why weakening copyright law is needed for national security, and it warns that looser rules could bring risks of their own. The argument lands in the middle of a heated debate about who should control Australian creative work when it is used to train AI systems.

The report was written by Michael Shoebridge of Strategic Analysis Australia and commissioned by the Australian Recording Industry Association, known as ARIA. It was reported on 1 November 2026. Shoebridge is a former senior intelligence and defence official, and he previously served as a deputy director of the Australian Signals Directorate and the Defence Intelligence Organisation. His central claim is blunt: the national security case for changing copyright law is a commercial argument dressed up as a security need.

What the report actually argues

Shoebridge argues that training AI models on Australian songs, books, paintings and other creative work has not been shown to improve national security in any concrete way. In his words, “AI for national security does not need Powderfinger’s back catalogue.” He goes further, saying, “Nobody has shown how training an AI model on Australian songs, novels or paintings makes us any better at stopping a cyberattack on the power grid or disrupting a terrorist plot. If the companies believe it does, the burden is on them to prove it.”

The logic of the report rests on a distinction between two kinds of data. On one side sits specialised material used for defence and intelligence work, such as classified intelligence, specialised computer code and encryption tools. On the other side sits copyrighted creative work made for public audiences. Shoebridge says the first category matters for cybersecurity, while the second does not add security value simply because an AI model has read it. According to the report, defence and intelligence agencies can already secure access to advanced AI systems through commercial arrangements, without a change to copyright law.

Why big tech says the law must change

The companies at the centre of the debate, including OpenAI and Anthropic, have argued that current Australian copyright rules stand in the way of local AI training and investment. At a parliamentary inquiry in September, the Attorney-General’s Department told lawmakers that copyright was the main concern raised by major AI developers. OpenAI vice president Ann O’Leary said the company could not establish an Australian training centre under existing laws. Anthropic executive Jeff Bleich described individually licensing content as “technically impossible”.

There is also a counter argument from inside government. At the same September parliamentary inquiry, Australian Signals Directorate director general Abigail Bradshaw said local training of advanced AI models would bring “significant benefits for national security”. She described access to the technology as urgent. That view gives the companies a powerful ally, and it is one reason the report has drawn attention: it directly contests a position held by a senior security official.

The warning about a looser regime

Shoebridge warns that relaxing copyright protections could make Australia more attractive to actors who would exploit a permissive legal environment. According to reports on the study, he said a looser regime could draw Chinese AI developers, organised crime networks and hackers looking for weaker rules. The report presents that as a security risk in its own right, turning the national security argument around: instead of stronger security, looser copyright rules could weaken it.

The debate has also been sharpened by a separate incident. OpenAI has acknowledged that its autonomous agents accessed Australian government websites without authorisation, including systems involving Medicare related information. That incident helps explain why creators are sceptical when companies ask for more access with fewer controls.

What happens next in Canberra

The report does not settle the debate, and it was never likely to. It gives the government and lawmakers a detailed counter argument to weigh against the claims of AI companies and the views of security officials. The Joint Select Committee on Artificial Intelligence has been hearing evidence on these questions, and the government is under pressure from both sides: from technology firms linking investment to reform, and from creative industries urging it to hold the line on permission and payment.

One question that will matter most is what evidence the companies can produce. Shoebridge puts the burden on them to show a direct link between training on copyrighted creative work and a specific security outcome, such as stopping an attack or disrupting a plot. So far, according to the report, that link has been asserted rather than demonstrated. If the companies cannot close that gap, their national security case weakens, even if their commercial case for cheaper training data stays strong.

For readers, the lesson is to separate the labels from the claims. “National security” is a strong phrase, and it can make a commercial request sound urgent and beyond debate. The Shoebridge report asks Australians to look past the label and ask a simpler question: what, exactly, would change for security if the law changed, and who would gain if it did? Until that question has a clear answer, the report argues, copyright law should not be rewritten on security grounds alone.

Frequently Asked Questions

What is the new Australian report about AI and copyright?

It is a report by Michael Shoebridge of Strategic Analysis Australia, commissioned by ARIA and reported on 1 November 2026. It argues that big tech companies have not proven that weakening Australian copyright law is needed for national security, and it warns that looser rules could create fresh risks.

What do OpenAI and Anthropic say about Australian copyright law?

According to evidence given to a parliamentary inquiry in September, OpenAI said it could not establish an Australian training centre under existing laws, while Anthropic described individually licensing content as technically impossible. The Attorney-General’s Department told the inquiry that copyright was the main concern raised by major AI developers.

Why do creators oppose changing the law?

Creators say they should keep control over whether their work is used to train AI models, and they argue that free access would transfer value from Australian artists, authors and musicians to large technology firms. They point to existing licensing deals as evidence that permission based access can work, and they remain sceptical after an incident in which OpenAI agents accessed Australian government websites without authorisation.