Welcome to Neural Notes, a weekly column looking at how AI is impacting Australia. In this edition, we explore how Australia's AI story in 2025 has been defined by rule-making rather than product drops…the majority of which arrived in the last few months of this year.
The Albanon government finally made some moves on regulation, creators pushed back against “free data” for training models, and some local startups raised significant funding.
Related Articles Block Placeholder
Article ID: 328096
Let's take a look back at some of the most important AI news of the year.
AI governance without AI law
Just a few weeks ago, the Australian government finally decided not to follow the EU's lead and enact its own AI legislation.
This was outlined in the long-awaited (some might argue “overdue”) National AI Plan. This frames Australia's approach to facilitating adoption and investment while managing risk through existing privacy, consumer, competition and workplace laws, rather than a separate AI Act.
The plan suggests greater collaboration between regulators, the creation of an AI safety institute, and targeted reforms where gaps exist. However, it did not lead to the introduction of new criminal offenses or a formal risk classification system.
This decision has real implications for Australian businesses. Domestically, many companies still broadly operate under the same legal frameworks they used before generative AI became mainstream.
Overseas, exporters are dealing with much stricter regulations. Australian SaaS companies selling to Europe are currently working on a phased rollout of the EU. AI lawThis includes conformity assessments, documentation requirements, and enhanced transparency obligations for high-risk applications and general-purpose models. In reality, this creates a fractured reality. Relatively flexible domestic regulations are combined with strict offshore compliance expectations.
Related Articles Block Placeholder
Article ID: 328397
Guardrails have certainly become stricter within the government. Updates to policies on the responsible use of AI in government have strengthened expectations around governance, risk assessment, and transparency, especially for high-impact applications.
For founders and vendors, this has led to closer scrutiny of procurement questionnaires and audit formats when selling AI tools to the public sector.
The subsequent decision to quietly abandon plans for an independent AI advisory body has only reinforced concerns about the extent to which external oversight will run alongside this self-regulatory approach.
Copyright, training data, and “no free lunch” decisions
The most important policy decision this year was not about the AI bill but about copyright law. After months of consultation and pressure from big tech companies and lobbyists alike, the federal government has ruled out a broad text and data mining exemption that would allow AI developers to train models on copyrighted books, journalism, images and music without a license.
Copyright has become an ethical fault line in Australia's AI debate, with ministers positioning the decision as protecting the rights of creators and ensuring fair compensation.
Related Articles Block Placeholder
Article ID: 255488
Of course, the training ship had already been sailing successfully for several years leading up to this decision.
Still, the choice forces difficult trade-offs for AI companies, especially those looking to train large models locally. Developers must either pay licensing fees for content, rely on more limited first-party or synthetic datasets, or move more model training offshore under more permissive regimes.
In the short term, that uncertainty has led many Australian startups to build their products on top of US and EU-based frontier models, rather than investing in training large-scale models from scratch.
For companies deploying AI tools, it also raises unanswered questions about liability, the provenance of data, and who is responsible when platforms shift risk back to users. And the risk-averse argument is not hypothetical. In Germany's GEMA case, OpenAI argued that the responsibility lies with the user who starts the system (a claim the court rejected).
Offshore rules, domestic impact
Internationally, 2025 was the year when AI regulation started to feel less theoretical and more practical. EU AI law Although the United States continued its gradual rollout, the United States relied on executive orders, agency guidance, and enforcement actions rather than a single national law. At least for now.
Still, US regulators and cloud providers are effectively setting global standards through requirements built into major platforms.
For Australian businesses, this has turned compliance into an infrastructure layer function. Hyperscalers and model providers currently sell their products as “EU”. AI law-ready” or in line with America’s new safety expectations.
As a result, power asymmetries are increasing. Australian founders are absorbing the cost and complexity of offshore compliance without having much say in how the rules are written.
For many midmarket startups, the strategic question in 2025 was not which model was the most powerful, but which vendor's compliance story would reassure customers, regulators, and boards of directors.
Related Articles Block Placeholder
Article ID: 329000
It’s a good year for AI startup funding, but there are some pitfalls.
Despite these regulatory complexities, 2025 was a strong year for AI funding in Australia. According to a report by Dealroom, the combined enterprise value of more than 470 Australian venture capital-backed AI startups is approximately US$11.7 billion (AU$17.6 billion). And this is not surprising. In our weekly startup funding roundup, 'AI-first' companies dominate in 2025.
Cut Through Ventures' Q1 data also pointed to the hype around AI funding. However, as we reported at the time, the data was somewhat skewed by the fact that many startups forced AI into their pitch materials to attract the attention of venture capitalists. And it worked.
One of this year's standout deals was Sydney-based healthtech company Harrison.ai's $179 million Series C, one of the largest AI-related fundings in Australian history.
Elsewhere, AI-applied companies in logistics, agriculture, and SaaS saw steady medium-term capital inflows.
Across markets, investor behavior has certainly shifted from generic chatbots to vertical, domain-specific AI products designed to connect directly to real-world business workflows and global markets.
How Australian businesses are leveraging AI in practice
For most Australian businesses, the AI story of 2025 has made AI boring. Companies are incorporating AI into their office suites, developer tools, and customer service platforms to change the way they write emails, resolve tickets, and distribute code.
This was supported by a series of frontier model releases from OpenAI, Google, and Anthropic, delivering improved multimodal inference, longer context windows, and tighter cloud integration.
OpenAI launched the affordable ChatGPT Go in India before rolling it out to select countries around the world, with market share also at stake. This “affordable” strategy is probably balanced by a less gradual transition to advertising on the platform.
Small and medium-sized enterprises adopted as well as large enterprises, but from a lower base and more unevenly. Research cited in Labor’s National AI Plan, based on research from the National AI Center and Quadrant 5, found that more than a third of Australian small and medium-sized businesses have implemented AI. That adoption rate fell to 29% in local organizations, compared to about 40% in metropolitan areas.
The plan also found that around a quarter of regional businesses are completely unaware of the AI opportunity.
When implementation did occur, it was often narrow and tool-driven rather than strategic. Small and medium-sized businesses relied on AI-assisted quoting tools, auto-generated marketing content, and chat-based administrative support that were often built within existing software rather than deployed as dedicated systems.
What has been lagging behind is practical support around governance, training, and accountability, leaving many small and medium-sized businesses moving forward with AI adoption with limited guidance at a time when voluntary disclosure and self-regulation remain key policy settings.
In other words, there is significant room for improvement in 2026. This means there is significant room for improvement, especially for those grappling with soft laws at home and tougher obligations abroad, while still relying heavily on global platforms to fill the gap.
