Regulating AI the right way
11 August 2025 | Stephen King and Julie Abramson
This article appeared in the The Australian on 10 August 2025.
History doesn’t repeat itself, but it often rhymes. For all that’s new about AI, the challenges and opportunities it presents are akin to those we’ve faced in the technological shifts of the past.
These technological shifts – from computers to the internet – are partly why the average Australian today has living standards three times higher than the average Australian in the 1960s.
In our recently released report we show just how great we think the opportunities from AI could be. No one can precisely predict how AI will change the economy, but our relatively conservative estimate puts the benefits to Australia at $116 billion over the next decade. That’s around $4300 per person.
The benefits of AI are significant but the potential harms can’t be ignored. Government needs to set the rules of the game to ensure that Australians reap the benefits of AI while mitigating the risks.
The question is not whether we protect ourselves from these harms but how we protect ourselves. Far from relaxing regulations, we should bolster them as necessary to address any gaps exposed by AI and ensure we have the tools and resources to enforce them.
But crucially, we should start with what we already have. Australia has robust laws and regulations in place that already protect us from fraud, discrimination and negligence to name a few. With measured reform that responds to any gaps that are identified, these laws can continue to safeguard us through the age of AI. We should only consider new AI laws if we’re certain that is no longer the case.
If we leap to economy-wide, AI-specific regulations without this process, we risk putting ourselves out of step with the rest of the world, creating uncertainty and limiting people’s willingness to adopt and invest in this transformative technology.
As in the early years of mass internet access, copyright deserves particular attention. In our inquiry, we heard from content creators who were concerned that AI developers are using their work without consent or compensation.
This is an important issue that deserves a considered approach. For this reason, we are seeking views from the public about whether there is a need for reforms to the copyright regime to meet any new challenges posed by AI technology. We have not reached a view or made any recommendations about what needs to be done.
One of the challenges is that the large AI models we’re all familiar with are trained in other countries, where Australian copyright laws do not apply. Any reforms Australia makes have limited influence over the way our copyrighted material is used by tech companies overseas.
What we can control is the regulations that govern how AI-training happens in Australia. And as it stands, many have told us that our approach to copyright has effectively taken Australia off the map when it comes to AI-training, including for certain small task-based models that might be used by research institutions or medical research firms.
That’s led some to advocate for a ‘fair dealing exception’ for text and data mining like those that apply in the EU or Singapore. Such an exception would not be a ‘blank cheque’ for all copyrighted materials to be used in AI models – the use may also need be considered ‘fair’ in the circumstances, and this would help preserve copyright holder’s interests.
Others have told us that while copyright laws are already sound, there are issues with tech companies sidestepping existing licensing and enforcement mechanisms. We have asked for feedback about whether there is a need to bolster the licensing or enforcement regimes. This could involve a greater role for collecting societies, who already act on behalf of artists, musicians and creators generally.
We are still collecting submissions on these questions before we submit our final report. However, our consultation so far suggests that the status quo is not working for content creators or those investing in the AI future.
This is just one of the many challenges that AI presents, but we have adapted to meet these challenges before. The scale of the AI revolution may well prove unprecedented, but so far many of the risks and opportunities it presents are familiar. Before we introduce blanket regulations, let’s be sure that the safeguards that have seen us through the technological shifts of the past can’t serve us through this one.
