Open gate policy?

Anthony Albanese’s announcement yesterday of a new national AI strategy comes after years of uncertainty about Australia’s approach.

Since the release of ChatGPT in late 2022, Australia has largely dithered while the AI hype train has relentlessly rolled on, with billions invested, mostly into a handful of outfits operated or owned by the usual big tech suspects. Several initiatives went nowhere, including the welcome early public consultation on Safe and Responsible AI, proposed mandatory guardrails, and the Artificial Intelligence Expert Group.

The new strategy relocates control to Prime Minister and Cabinet, signalling a coordinated approach to AI, or more cynically, a single door for lobbyists to come knocking. While the Prime Minister’s speech was long on rhetoric and short on detail, the rhetoric clearly signalled a government willing to take control over the future of AI in Australia. There was talk of sovereignty, of national opportunity and public benefit to accompany the inevitable waves of big tech investment. There was also reassurance on risks to the environment, to jobs, and to the sustainability of the creative and media industries.

In the clearest statement yet on the last of these, Albanese said, “No company should use Australian books, music, art or news to build or train AI without the artist’s control. That includes the artist’s control of the price and value of their work. Anything less, is theft.” On the one hand, this might provide the policy certainty which big tech has been clamouring for. On the other, there’s no doubt it’s not the policy big tech wants.

And the details are still missing.

Like the many-eyed Irene, devourer of worlds, AI companies have vacuumed up everything they can find to feed the models they are now deploying for commercial gain. News companies, journalists and other creators have already lost control of their published work, and thus their ability to benefit financially from it. They may be able to claw some value back through private deal-making and licensing schemes, with clarity on copyright potentially boosting their bargaining power. But media’s loss of control over content goes beyond access and commercial value.

AI increases gatekeeping power of tech companies over news and other digital content—a power that has already driven the incredible growth of big tech over the past two decades. This was one of the most serious concerns voiced by the news editors we interviewed in our research on AI and journalism. It represents a loss not only of commercial value, but of editorial control. The impact of this loss extends beyond the private interests of media companies to the interests of the public in the availability of reliable and diverse sources of news and information. Neither private deal-making nor licensing schemes necessarily address this impact, since they do not stand in the way of the gatekeeping power of AI.  The biggest beneficiaries of these schemes are likely to be larger media companies.

There is tremendous benefit in having news and other reliable information available and accessible, at no cost, to the public. For this reason, governments see fit to fund public information infrastructure, including public libraries and public broadcasters. Some argue that the same logic should lead us toward the funding of public AI; even more so, given AI’s reliance on public-domain and publicly accessible works.

There is no sign in Albanese’s speech that we are heading down that path. And with AI explicitly excluded from the revamped News Bargaining Incentive scheme, which in its original form at least sought to address some aspects of the gatekeeping power of digital platforms, it remains uncertain how the impacts of that power on the public information ecosystem will be addressed.