Newsroom
AI is already in your broking business. Is it governed?
- Oxcel Finance
- October 8, 2026

Most Australian brokers now use AI in some form, but survey data shows many do so without a written policy or guardrails. ASIC's new review of AI use by banks, announced on 30 September 2026, signals where regulatory attention is heading. The brokers who benefit most from AI will be those who can show their clients, lenders and regulator that they use it safely.
Adoption has outrun governance
Broker AI adoption has moved quickly. Connective's AI Readiness in Australian Broking Report, published earlier in 2026 and based on a survey of more than 300 mortgage, commercial and asset finance brokers, found 86% believe AI will be essential or helpful to their business over the next two years.
The same research found 37% of brokers use AI tools regularly and confidently, and around 14% had not adopted any AI. General-purpose large language models are the most common tools.

The gap is governance. Commentators reviewing the Connective findings noted many brokers use AI without policies or a clear understanding of the risks. Connective itself recommends a Responsible Use of AI policy, accuracy checks and aggregator guardrails.
Regulators are moving from interest to review
ASIC's 2026–27 banking priorities letter commits it to a banking-sector AI review, expected to begin in Q2 2026–27 (October–December 2026), focused on new and proposed AI use cases and their impact on customers. ASIC will coordinate with APRA.
The letter references ASIC's Report 798, published in October 2024, which found AI adoption across financial services was outpacing governance and consumer-risk assessment. ASIC says it wants consumer protections maintained in customer-facing interactions, decision-making and lending processes.
The review targets banks, not brokers. But lenders' AI governance will flow through to how they accept broker-submitted applications, documents and data. And brokers remain fully accountable for best interests duty outcomes, whatever tools they use.
The four real risks for brokers
Privacy. Pasting client payslips, bank statements or identity documents into a public AI tool may breach the Privacy Act and your licence obligations. The question is where the data goes, who can access it and whether it is used to train models.
Accuracy. Generative AI can produce confident, wrong answers about lender policy, serviceability or product features. An inaccurate output repeated to a client is the broker's error, not the tool's.
Best interests duty. AI can help shortlist products, but the broker must still be able to explain why a recommendation is in the client's best interests. "The tool suggested it" is not a BID file note.
Fraud. The same technology that saves time also makes fake documents easier to produce. AUSTRAC's Operation Claw findings involved fake payslips and invoices. Lenders are investing in AI-based fraud detection, and broker files will increasingly be scanned by it.
Where AI genuinely helps
Used well, AI can reduce administrative load in areas such as:
- Drafting client emails and summaries, reviewed by the broker before sending.
- Structuring fact-find notes and BID file notes from the broker's own input.
- Checking documents for completeness before submission.
- Summarising lender policy updates from official sources.
- Monitoring the loan book for repricing and fixed-rate expiry triggers.
The common factor is that a human reviews the output and owns the decision.
Key facts
- 86% of brokers surveyed by Connective see AI as essential or helpful over the next two years.
- 37% use AI regularly and confidently; around 14% have not adopted AI.
- ASIC banking-sector AI review: expected Q2 2026–27.
- ASIC REP 798 (October 2024) found AI governance lagged adoption.
- Scams Prevention Framework scheduled to commence March 2027.
What this means for brokers
AI is becoming a baseline productivity tool. Brokers who ignore it will carry higher costs per file. But brokers who adopt it carelessly carry compliance and reputational risk that can outweigh the time saved.
Lenders and aggregators will increasingly expect evidence of controls. A written AI policy, approved tools and staff training are likely to become standard expectations, much as cyber security controls have.
The human relationship becomes more valuable, not less. If routine tasks are automated, the broker's value is judgement, structuring, negotiation and accountability.
What brokers should do now
- Write a one-page AI use policy covering approved tools, prohibited data and mandatory human review.
- Never upload identifiable client documents to tools that are not approved by your licensee or aggregator.
- Require a human check of every AI-generated client communication and file note.
- Ask your aggregator which AI tools are approved and how client data is protected.
- Train staff to spot AI-generated document fraud, including inconsistent payslips and invoices.
Oxcel perspective
Technology should give brokers more time with clients, not replace the judgement clients pay for. We are in the people business first, finance second. AI should make that easier, with governance built in from the start.
General information only. This content does not constitute financial, credit, legal or tax advice. Finance products and lending criteria are subject to lender assessment and eligibility requirements.
Sources & further reading
- ASIC — Supervisory letter: 2026–27 banking sector priorities, 30 September 2026.
- Connective — AI Readiness in Australian Broking Report, 2026.
- Australian Broker — Brokers back AI but most still flying blind on strategy, 2026.
- The Adviser — Brokers think AI will benefit their business but lack strategy, 2026.
- AUSTRAC — Fintel Alliance uncovers coordinated mortgage fraud across major lenders, August 2026.






