Australian firms rush AI adoption amid rising risks

Australian businesses are accelerating investment in artificial intelligence while grappling with growing governance, security and operational risks. Commentators across technology, HR, retail and cybersecurity say decisions made in the next few months will shape how resilient organisations are as AI and automation spread.

AI agents that act on business systems are sharpening long-standing concerns about data sovereignty and accountability. Early deployments focused on tools that answered questions or drafted content. A new wave of systems now initiates changes, places orders and routes work, shifting the risk profile for boards and regulators.

Dion Williams, Founder and Chief Executive Officer at Australian workflow vendor Servicely, said AI that can take action is forcing a different discussion inside local enterprises.

"Once you let AI agents take action rather than just answer questions, governance stops being a compliance checkbox and becomes central to the design. Where does the data live? Who is accountable when an agent acts? What keeps working if a foreign provider changes the rules? Australian enterprises increasingly want answers to those questions before they scale, and a locally accountable platform is part of the answer," said Williams, Founder and CEO, Servicely.

The governance debate now extends beyond IT departments. HR leaders say boards are underestimating AI's impact on the workforce and decision-making, even as adoption outpaces risk frameworks.

HR specialist Evelina Samuels said organisations often treat AI oversight as a technical exercise rather than a people and culture issue.

"Many organisations have moved quickly to adopt AI to improve productivity, while governance is still playing catch-up. But one of the biggest misconceptions is that AI governance is purely a technology issue. AI isn't just changing systems, it's changing how people work, make decisions and build capability. That's why I think AI governance is becoming a people issue just as much as a technology issue," said Samuels, HR Expert, Samuels Doneghan.

Retailers face the same AI and data challenges while confronting rising wage costs, volatile demand and ongoing store closures. Infios Director of Solution Delivery Darren O'Connor said the sector is under pressure to connect fragmented systems and data across stores, warehouses and transport.

"The wave of retail closures we're seeing in Australia reflects changing consumer behaviour, but it also highlights the increasing complexity retailers are managing behind the scenes. As disruption, from geopolitical uncertainty to shifting market dynamics, continues to accelerate, businesses need intelligent fulfilment operations that can adapt quickly, make better decisions and execute with greater precision. Those that connect their operations end to end will be better positioned to compete and grow," said O'Connor.

He linked the latest minimum wage rise to a renewed focus on productivity across the fulfilment chain.

"With the minimum wage increase taking effect in July, retailers are facing further cost pressure at a time when they are already navigating significant market challenges. As operating costs continue to rise, the focus must shift to improving productivity and eliminating unnecessary complexity across the fulfilment lifecycle. By connecting people, processes and technology through intelligent supply chain execution, retailers can reduce manual effort, make faster decisions and build the agility needed to adapt in a changing market," O'Connor said.

Retailers that rely on disconnected order, inventory and logistics systems struggle to deliver consistent omnichannel service, he said.

"Meeting today's customer expectations requires more than operating across multiple sales channels - it requires a fulfilment operation built to connect and execute seamlessly. Omnichannel success depends on intelligent supply chain execution, where orders, inventory, warehousing and transportation work together as one. This enables retailers to respond faster, make better decisions and deliver a more consistent customer experience," he said.

AI-driven visibility is only as effective as the data flows between teams and systems, O'Connor said.

"Retailers today have more data at their fingertips than ever before, but the challenge is turning that information into timely action. When systems are disconnected and decisions rely on manual intervention, opportunities to respond quickly are lost. Intelligent supply chain execution, enabled by AI, helps connect people, processes and technology to provide greater visibility, improve decision-making and orchestrate more agile fulfilment operations," he said.

He warned that the second half of the year leaves a shrinking window for structural fixes before peak demand.

"Retailers in Australia are facing a more competitive and unpredictable operating environment, where consistency of service and speed of response are key differentiators. Meeting those expectations requires intelligent supply chain execution: connecting inventory, capacity, cost and customer commitments in a single, coordinated environment, and using data and AI to enable faster, more confident decisions.

Retailers heading into the second half of the year have very little runway to fix structural issues in their supply chain before peak season arrives. The window for meaningful preparation is narrowing faster than many businesses realise, and the choices made now, from inventory positioning to order orchestration and fulfilment execution, will shape how resilient operations are when demand peaks later in the year," he said.

In cybersecurity, pressure from new large-scale AI models is raising similar questions about how far automation can go. Some vendors and investors now promote fully automated security operations centres that use AI to triage alerts and respond without human intervention.

ThreatDown Cybersecurity Evangelist Mark Stockley said that approach risks both operational failure and spiralling infrastructure costs.

"Replacing human defenders with an AI SOC is a multi-million-dollar trap. By now we've all used AI enough to know that it's often brilliant and occasionally awful, and you don't get to decide which version you get on any given day. You wouldn't hire a human security analyst who behaved like that, so why accept an automated one? Between astronomical token costs and the simple fact that automation isn't ready to stand alone, enterprises are setting themselves up for failure. Organisations have perhaps six months to reassess their security posture and prepare for Mythos-class AI models causing a deluge of urgent security patches and a torrent of new exploits. An effective response isn't to hand everything over to AI, but to find the best combination of AI speed and human judgment," said Stockley.