Australian SME AI Adoption: What the Latest National AI Centre Data Shows

The headline figures

The National AI Centre’s AI adoption insights for December 2025 to February 2026 were first published on 7 May 2026 and display an updated date of 3 June 2026. The report says 43% of Australian SMEs reported some level of AI adoption across the quarter, compared with 45% in the previous quarter. It also says adoption rebounded to 44% in February 2026 after a slight dip in January.

The figures come from the SME AI Pulse, a monthly survey commissioned by the National AI Centre and conducted by Fifth Quadrant. According to the report, each wave surveys at least 400 SME owners and decision-makers and is weighted by industry, state and employee size. “Some level of AI adoption” is not the same as integrated workflows or measured business benefit.

Trust is the main reported barrier

Among non-adopting businesses that do not intend to implement AI in the next 12 months, the report says around 65% cited distrust in AI decision-making or a strong preference for human control. Those percentages describe that subgroup, not all SMEs.

Our interpretation is that “human control” should be treated as a design requirement. Businesses can use AI to prepare, organise or recommend while preserving review for decisions that affect customers, money, staff or compliance. Clear explanations of limits are often more persuasive than claims of autonomy.

Relevance and know-how matter

The report says 54% of non-adopting businesses considered AI not relevant to their business. It also says 19% of SMEs reported not knowing how to use AI in their business, up two points from the previous quarter. The report highlights lower adoption in construction and agriculture and higher adoption in health, education and services sectors.

A relevance problem may reflect a lack of specific examples. A construction firm, accountant and retailer have different workflows, risks and systems. Practical guidance should begin with their actual handovers and information, not generic productivity promises.

Adopters are going deeper

The report says broad adoption reached its highest level in seven months while limited one-off use declined. Among businesses using or planning to use AI, content generation and data analytics were each reported by 54%, followed by cybersecurity and threat detection at 48%. More complex applications, including agentic AI, remained largely untapped according to the report.

The National AI Centre also reports that checking outputs before they affect customers was the most common responsible-AI safeguard, used by around half of current users. Customer transparency and processes for raising concerns lagged. That gap is a practical issue for businesses deploying customer-facing AI.

What the data cannot tell us

The report does not prove that adopting AI will improve profit or productivity for a particular business. It measures reported adoption and attitudes. A one-point monthly increase also should not be overstated as a durable trend without wider statistical context.

For an individual business, the better question is whether a specific workflow can be improved safely and measurably. Use the adoption data as context, not as pressure to buy technology.

What this means for Australian businesses

Australian SMBs should not assume they are behind simply because a competitor uses AI, nor dismiss it because a first experiment felt irrelevant. Focus on recurring work and clear safeguards.

Customer-facing use should include output checks, appropriate disclosure where relevant and an accessible way for customers to raise concerns.

Practical next steps

1. Identify a recurring workflow specific to your industry.
2. Define where people must review or approve outputs.
3. Start with a contained, low-risk use case.
4. Tell staff and, where relevant, customers how AI is used.
5. Measure local results instead of relying on adoption headlines.

See MindFish’s implementation process