Open AI has broken ranks with the Tech Council of Australia over copyright restrictions, declaring that its artificial intelligence models are “going to be in Australia, one way or the other”.
Chris Lehane, the chief global affairs officer of the artificial intelligence company responsible for ChatGPT, made a keynote appearance at SXSW Sydney on Friday, where he covered the geopolitics of AI, Australia’s tech future – and the global debate around using copyrighted material to train large language models.
Scott Farquhar, the Tech Council chief executive and Atlassian co-founder, has previously said Australian copyright law “hurts a lot of investment of these companies in Australia”.
Sign up: AU Breaking News email
In August, it emerged the Productivity Commission was examining whether technology firms should be exempt from copyright rules that stop companies mining text and data to train AI models.
But, when asked whether Australia could lose investment in AI development and datacentres if it didn’t loosen its copyright laws to fair use, Lehane told the audience:
“No … we are going to be in Australia, one way or the other.”
Lehane said countries generally chose one of two positions when it came to copyright restrictions and AI. One was to take a US-style fair use approach to copyright, allowing for the development of “frontier” (highly advanced, large-scale) AI, while the other was to maintain a historic position on copyright, limiting AI’s scope.
“We will engage in either country – we will find ways to work with those who want to build up big frontier models and have robust ecosystems, or those who just want to have much more narrowly defined AI,” he said. “We will work with them under either scenario, regardless.”
When questioned about Sora 2 – Open AI’s new video-generating model – being launched and monetised before copyright use had been ironed out, he said the company was benefiting “everyone”.
“This is the nature of how technology works. Innovations come along, and then societies adapt to those innovations,” he said. “We are a non-profit, and our purpose is to build AI that benefits everyone … in the same way that a couple of generations ago, people could go to the library and learn.”
Open AI on Friday suspended the ability to generate videos featuring likenesses of Martin Luther King Jr after his family complained about the technology.
Lehane also said China and the US were engaged in a “very real competition” to shape the future of global AI, pitting values head-to-head.
“We don’t necessarily think about this as a fight as much as it is a competition – but it is a very real competition, and the stakes are pretty significant,” he said, adding US-led frontier models will “inherently be built on democratic values”, while those of China will “probably” be built on autocratic norms.
“One of the two will end up being the player that the rest of the world builds on,” he said.
When asked whether he was confident that America would remain a democracy, he said, “democracy, as others have said, tends to be a very messy process – but the US has demonstrated over time that it gets this stuff right.”
US and its allies – Australia included – are going to have to start generating a gigawatt of energy on a weekly basis to develop the infrastructure needed to maintain a “democratic lead” on AI, while Australia has the potential to develop its own frontier AI, he added.
“Australia occupies a really unique position” in that it has a very high AI user base, 30,000 developers, “a tonne of talent”, a fast-growing renewable sector, fibre optics connections with Asia and is a Five Eyes nation, he said.
