Sovereign AI model poised to become Australia’s ChatGPT – Information Age | ACS
Thanks for signing up!
Sorry there was an error with your request.
By David Braue on Sep 09 2025 12:42 PM
The compensation of copyright holders for AI training has been prioritised as a new “ethics-driven” Australian AI model, designed as a fully sovereign alternative to ChatGPT, debuts on the back of an investment in Australia’s largest AI supercomputer.
Newly launched Sovereign Australia AI has ordered 256 Nvidia Blackwell B200 GPUs – which when installed will provide even more raw computing capacity than the record-setting H200 based AI supercomputer announced last month by ResetData.
The B200 chips include 208 billion transistors – 2.6 times as many as the H200 – and consume more power, delivering almost twice as much memory bandwidth in an AI architecture that co-founder and CEO Simon Kriss lauded as game-changing.
"We’ve secured the largest cluster of NVIDIA Blackwell GPUs in Australia,” he said, giving us “the compute power to train foundational models at a scale never seen here before, putting us on the global AI map."
The system will be hosted in Australian data centres operated by Australian data centre operator NEXTDC, with compliance to Australian privacy and security standards key as Sovereign Australia AI builds a locally trained large language model (LLM).
Acknowledging the “significant technological challenge” of building a large-scale generative AI (genAI) system, co-founder and chief technology officer Dr Troy Neilson downplayed suggestions the effort is an effort to directly compete with ChatGPT.
Instead, he said, the new LLMs – to be called ‘Ginan’ and ‘Australis’ and designed “to reflect the culture, language, and principles of this nation…. will serve as viable alternatives that better capture the Australian voice.”
Transparency is core to Sovereign Australia AI’s mission statement, with a commitment to openly share information about the data used to train its models and plans to open-source the Ginan research model for free public use.
Backing this commitment, the company has earmarked at least $10 million for the rights to content to train the systems – ensuring Australian copyright holders are compensated for their role in the training of the AI with “ethically sourced data sets.”
The commitment comes days after global AI giant Anthropic agreed to pay $2.3 billion ($US1.5 billion) to settle a class action brought against it by writers and publishers whose content the company appropriated to train its Claude AI system.
A June court ruling found that Anthropic’s use of the materials was legitimate ‘fair use’ under copyright laws, but that Anthropic had nonetheless violated copyright laws because of the way it acquired the content from pirated content websites.
OpenAI has previously argued that training genAI models is “impossible” without using copyrighted content – but by addressing copyright concerns from Day 1, Sovereign Australia AI is working to avoid similar issues.
Sovereign AI founders Dr Troy Neilson [L] and Simon Kriss are getting on the front foot with copyright holders. Photo: Supplied.
It’s a philosophical approach that has helped the venture secure the backing of the Australian Computer Society (ACS), the peak body of Australia’s ICT industry, and the publisher of Information Age.
The new system will “help ensure Australia’s technology workforce has access to sovereign, ethical AI built with our nation’s values at its core,” said ACS CEO Josh Griggs in welcoming the organisation’s collaboration with Sovereign Australia AI.
Delivery of a large-scale sovereign AI capability answers calls business leaders were making five years ago, years before OpenAI’s release of ChatGPT changed the way the world engages with information, and with each other.
The dominance of overseas tech giants – which are spending trillions on cloud and AI infrastructure in Australia and elsewhere – has prompted increased data sovereignty, procurement reviews, new satellites and local investment in sovereign data centres.
The early support of ACS validates Sovereign Australia AI’s mission to build the new system as a national asset, Kriss said, “and shows there’s a real appetite for sovereign, ethical AI solutions built right here in Australia.”
Sovereign AI solutions have become even more important, he said, given that “we are already seeing how AI is shaping the way people think, work, and engage with information.”
“If the foundational AI models Australians rely on are built offshore,” he said, “we risk losing control over how our national values are represented…. If we want models that reflect our values, protect our data, and build trust, we must have sovereign capability.”
David Braue is an award-winning technology journalist who has covered Australia’s technology industry since 1995. A lifelong technophile, he has written and edited content for a broad range of audiences across myriad consumer and business topics, with a particular focus on managing the intersection of technological innovation and business transformation. He has twice won Best IT Journalist at the Australian IT Journalism awards, and was named Best Technology Journalist at the 2024 Australian Technologies Competition.
