Defence report warns national security claims mask copyright concerns

A new defence report from Strategic Analysis Australia warns that national security claims are being used to justify weakening Australia’s copyright laws, potentially opening the country to cyber threats.

The report, commissioned by the Australian Recording Industry Association, criticizes lobbying efforts by companies such as OpenAI and Anthropic, which claim they cannot train advanced AI models in Australia without relaxing copyright protections.

Australian officials have faced pressure from Silicon Valley to allow AI developers greater access to copyrighted material, with tech giants linking major investments in Australian infrastructure to legal changes.

Former intelligence official Michael Shoebridge, who authored the report, argues that companies are using national security as a cover for commercial interests, claiming AI training on creative works is unnecessary for cybersecurity or counterterrorism.

Shoebridge noted that advanced cybersecurity relies on specialized code, encryption, and classified intelligence, not copyrighted content, and warned that weakening copyright could attract foreign AI developers and organized crime.

The report also raised concerns about the security credentials of major AI developers after OpenAI admitted its systems accessed Australian government websites without authorization, including systems involving Medicare data.

Shoebridge argued that Australia’s existing secure infrastructure and intelligence ties with the U.S. give the government leverage in negotiations, and that training American-owned AI models in Australia would not grant sovereign control over the technology.