As a tech reporter, I’ve spent years arguing internet access should be treated like a utility, particularly in a country that is still suffering from a digital divide.  Connectivity shouldn’t be treated as a luxury or a lifestyle perk. If you’re determined to weave technology into the basic fabric of society – from workplaces to schools to how we pay out bills – it needs to be affordable and accessible. Related Article Block Placeholder Article ID: 341054 Albanese grabs Australia’s AI agenda, parks copyright and data centre decisions Tegan Jones 4 And it seems generative AI may just end up sitting within that same category.  In a few short years, it has been folded into search, customer service, office work and coding. We’ve been sold the narrative that everyone must use AI, or they’ll be left behind.  And it’s gotten to a point where countries not only need to think about guardrails, AI policies and sovereign capabilities, but where the Australian Prime Minister has folded governmental AI direction into his own office. Smarter business news. Straight to your inbox. For startup founders, small businesses and leaders. Build sharper instincts and better strategy by learning from Australia’s smartest business minds. Sign up for free. * indicates required Email Address * By continuing, you agree to our Terms & Conditions and Privacy Policy. And that brings us to the issue of control. As AI becomes increasingly embedded into businesses and everyday life, the default models matter a great deal. If AI is indeed critical infrastructure, why are we defaulting to offshore — often privately owned big tech models — with dubious training practices? Current’s public option Related Article Block Placeholder Article ID: 339923 Neural Notes: The US keeps proving it can switch off Australia’s AI Tegan Jones Enter Current AI. This not-for-profit says it’s trying to do something frontier labs are not built to do: make AI feel like shared infrastructure rather than a private product. Current is building a public option for AI through a public-private partnership, backed by governments, foundations and other funders rather than investors.  The point is not another chatbot or subscription tier. Its model is built around open tools, community-led governance and systems meant to serve the public interest. Its early projects have been a testament to that vision. One is Suno Sutra, a pocket-sized offline device that runs AI in 22 Indian languages, built with Bhashini, the Indian government’s AI language division.  Another is AlphaChat, an open-source chatbot assembled in seven weeks by a coalition of 10 organisations, including Hugging Face, Mozilla and MIT Media Lab. Current has also allocated US$3.2 million in grants across projects in Kenya, Lebanon and the Brazilian Amazon. Its partnership with Tokyo-based Sakana AI suggests this is not just one non-profit experiment. Sakana has been pushing a different kind of AI stack for a few years as well. It’s building a model that puts more emphasis on local language, cultural fit, and a path well outside the Bay Area default. This partnership is interesting because it suggests a potential future with AI systems that are more regional, open, and less dependent on the same handful of businesses setting the terms for everyone else. Maybe we can actually build AI on terms that aren’t focused entirely on ROI – particularly when OpenAI is still firmly in the red. Related Article Block Placeholder Article ID: 298307 What is Sakana AI and why is everyone comparing it to OpenAI? Tegan Jones We could have a choice The existence of Current AI doesn’t necessarily mean it’s the solution. But it is a reminder that there could be alternative AI pathways. So far, all of the major players operate firmly in the for-profit enterprise space – even when they have gotten away with training on other people’s work.   Now you might argue OpenAI began as a non-profit. However, it moved into a capped-profit structure so it could raise the money frontier AI needs. Since then, it has kept drifting closer to a conventional for-profit model.  Anthropic took the route of public-benefit corporation language and safety branding. But it is still a proprietary lab. It still depends on commercial control, closed systems, and growth pressure. As for Meta, Google and Microsoft – they’re all publicly traded companies with shareholders to placate. And Anthropic and OpenAI aren’t far behind this same fate. The other thing all of these businesses have in common is positioning AI as imperative infrastructure for the future of, well, just about everything. For most of the past few years now, AI has been injected so heavily into the digital stack that the obvious intent is to make it not just a platform but the plumbing itself.  Eventually the term itself will likely become redundant as businesses, governments and individuals rely on it without even thinking about it.  The awkward part about that is the current dominant systems weren’t built for neutrality. We’ve seen that with the scramble for market share. So what we’re left with now are questions around ownership, cost, access and control. Anthropic’s US$1.5 billion copyright settlement this week is a great example of this. The case centred on allegations that it downloaded and stored millions of copyrighted books, including material from pirate sites.  And it’s not the only one. OpenAI, Meta and Google have all faced similar cases. It’s become a familiar pattern — huge datasets built with little to no consent and a legal fight after the fact. Related Article Block Placeholder Article ID: 339545 We grew up fearing piracy laws, now AI is rewriting them in Australia Tegan Jones But it doesn’t change the origin of these platforms or that they have been largely allowed to get away with building on stolen content to the point where governments are retroactively looking at licensing deals years after the train left the station. Even if AI is unavoidable at this point, are we really happy to allow the default to be the one built on theft – which we’re then expected to pay for? But let’s park the ethical issues for a moment and look at the for-profit model. In theory, there’s nothing inherently bad about this. Making money for goods and services is what private companies do. However, if you’re pushing something towards the operating layer of almost every facet of society, that has to change the conversation. If something is essential, then access matters. If access matters, governance matters, and if governance matters, ownership matters. This is especially true in Australia, where the questions around sovereign capability, copyright, model access and dependence on foreign infrastructure are all landing at once. Not to mention the debate around data centres popping up all over regional Australia. Australia is also, slowly, building the policy machinery around this. The national framework for the assurance of AI in government, along with newer federal AI policy, treats governance, procurement, risk and accountability as imperative. Related Article Block Placeholder Article ID: 322045 Neural Notes: We’ve let ‘AI innovation’ become synonymous with theft Tegan Jones Importantly, this is happening around yet another major telco outage, which has seen Telstra hauled in front of a senate committee because its infrastructure failed. A major investigation is also ongoing. And that’s exactly as it should be, and I daresay what will need to happen with AI. If AI is here to stay, it’s worth asking whether its default model should look more like public infrastructure than an increasingly pricey private paywall. And if we do indeed become reliant on it and something goes wrong, will those in charge be held to the same level of accountability as other service providers? That doesn’t have to mean state-run AI, or some utopian rejection of commercial systems. It just means the default shouldn’t be assumed.  If AI becomes unavoidable, then who it serves, who can afford it, and who gets locked out should be integral to the conversation. Neural Notes is a weekly AI column that looks at how AI is influencing Australia. Stay in the know Never miss a story: sign up to SmartCompany’s free daily newsletter and find our best stories on LinkedIn.