Australia’s insurers are deploying generative AI in claims at a rate that outpaces global peers, yet a widening gap between adoption speed and governance depth carries direct, practical consequences for the brokers who sit between carriers and their clients – and who, by their own admission, are not fully prepared.
According to Gallagher Bassett’s (GB) The Carrier Perspective: 2026 Claims Insights report, 74% of Australian insurers are now using generative AI in claims resolution, against a global adoption rate of 68%. Fraud detection, claims triage, and document management are the primary use cases, as carriers respond to rising claims volumes and growing complexity. The same report found that 70% of Australian insurers identify regulatory and compliance risk as the biggest barrier to further AI integration.
That figure is consistent with findings from CSIRO and the Insurance Council of Australia (ICA), whose joint 2025 report, AI for Better Insurance: Enhancing Customer Outcomes amid Industry Challenges, identified automated claims processing and triage, fraud detection, and enhanced underwriting and risk assessment as priority AI use cases, while warning that adoption must be matched with stronger governance. ICA CEO Andrew Hall said at the time that “the industry is committed to prioritising safe adoption by addressing privacy and safety concerns and AI system biases so that these technologies serve all Australians fairly.”
The governance gap, however, is not exclusive to carriers. Research released by the National Insurance Brokers Association (NIBA) at its 2025 Convention found that while 83% of brokers expect technology and automation to have significant impact by 2035, only 61% feel prepared – a 22-percentage-point preparedness deficit that makes AI governance at the carrier level a risk brokers may be structurally ill-equipped to interrogate.
A survey of 250 CFOs and senior finance leaders in Australia, published by Avalara in July 2026, found that 18% of respondents said that if a significant AI agent error occurred, accountability would either be unclear or belong to no one. In claims, that ambiguity has direct broker implications: where an AI-assisted decision produces a denial, a delay, or an incorrect assessment, clients will look to their broker to explain what happened and who is responsible.
The complaints data makes the exposure concrete. The Australian Financial Complaints Authority (AFCA) received 111,373 complaints in the 2025 calendar year – a 14% increase on 2024 and the highest annual volume in the authority’s history – with delay in claim handling the most complained-about issue across all financial services. As insurers increasingly explore AI-enabled claims processes, complaint trends in motor and home insurance highlight why governance, transparency, and human oversight remain key considerations.
Claims and complaint handling failures by insurers are among the Australian Securities and Investments Commission’s (ASIC) confirmed 2026 enforcement priorities, alongside misleading pricing practices and financial reporting misconduct. Automated decision-making is also set to come under formal Privacy Act regulation by December 2026, adding a further compliance dimension brokers will need to understand and communicate to clients at placement and renewal.
Gallagher Bassett’s report found that 90% of Australian insurers maintain in-house validation teams – the highest proportion across all regions surveyed – while 68% conduct manual review processes and 20% engage external verification services. Cammaron Blanchard, Gallagher Bassett’s head of operations automation and AI, said the role of those teams extends beyond risk containment. “Human validation of outputs is essential, especially at this crucial stage where organisations are scaling their AI capabilities. Oversights do more than manage risk. They’re an additional layer of quality control and should answer the fundamental question: Are your implementations delivering quality outcomes?” Blanchard said.
Blanchard said governance considerations now run through client relationships directly, with the organisation working with carriers to formalise AI approvals and contractual oversight provisions. “It’s important to consider that concerns around AI extend outside of the industry and into the consumer space. Because of this, I’ve worked closely with some of our major clients to implement AI approvals and contract provisions. The invitation to collaborate in this rapidly evolving space gives them confidence in what we’re deploying and how we’ll be managing it,” Blanchard said.
Whether such provisions exist – and what they cover – is a due diligence question that is becoming relevant for brokers at placement. No Australian industry body has yet published formal guidance for intermediaries on how to assess a carrier’s AI governance posture, leaving brokers to navigate that question without a standardised framework.
In its April 2026 letter to industry, the Australian Prudential Regulation Authority (APRA) warned that governance, risk management, assurance, and operational resilience practices are not keeping pace with the scale, speed, and complexity of AI adoption. APRA also observed an overreliance on vendor presentations and summaries without sufficient examination of key AI risks. APRA’s CPS 230 operational risk management standard, which commenced on July 1, 2025, requires APRA-regulated entities to identify, assess, and manage operational risks, including those arising from technology systems and material service providers. While CPS 230 does not specifically regulate AI, its requirements are relevant to insurers using AI in critical operations, particularly where they rely on technology providers or automated systems that may affect operational resilience and customer outcomes.
Blanchard said that governance and innovation are not in conflict. “Integrating good governance doesn’t mean sacrificing innovation; it begins with having the right conversations. In my discussions at GB, I’m positioning AI as an augmentation to what our team can do. It’s about using technology to support them and change the way they work. Everything has a human in the loop. Then our teams can focus on the exceptional judgement, empathy, and technical skill they bring to every claim,” Blanchard said.
For brokers, that framing offers a practical test to apply to their insurer and TPA relationships: not whether AI is being used in claims, but whether the human accountability layer behind it is demonstrable – and what recourse exists when it is not.
