Unlicensed websites with no Australian Financial Services Licences (AFSLs) were cited thousands of times in AI-generated insurance search results used by Australian consumers, according to new research – published four months after the Australian Securities and Investments Commission (ASIC) warned publicly that AI tools used for financial guidance carry important limitations that could lead to inaccurate or inappropriate suggestions.
Somantra, a Sydney-based AI search brand visibility platform, published “The AI Search Anomaly: How 38 Domains Contaminated Insurance Advice Across AI Search Engines” on July 21, 2026. The study examined 2,437,107 citation records across 28,725 unique domains drawn from Google AI Overviews and ChatGPT’s standard search feature between November 2025 and July 2026.
Of the domains reviewed, 38 were verified through a multi-stage process as confirmed spam or grey-area “parasite SEO” operators – sites constructed to simulate legitimate financial guidance while generating revenue through AI citation. None held an AFSL, maintained a verifiable Australian business presence, or bore any accountability for the content they published. The disparity between platforms was pronounced. Grey-area and spam domains accounted for 1.97% of all ChatGPT citations in the dataset, compared with 0.10% of Google citations – a nineteen-fold difference
One flagged domain alone was cited 5,366 times in ChatGPT’s Australian insurance-related responses, briefly ranking it as the 13th most-cited source in that category. The report identified a cluster of six domains sharing naming conventions that combined trust-signalling terms with insurance-adjacent language, all peaking simultaneously in January 2026 during Australia’s insurance renewal season – a pattern Somantra described as consistent with a single operator or coordinated group.
The findings land in context that ASIC has been actively shaping. In March 2026, ASIC’s Moneysmart program published its first-ever consumer guidance on the use of AI for financial decisions, responding to its own research showing that 64% of Gen Z Australians trust AI platforms for money advice and 18% actively use them for financial information and guidance. The guidance stated that while general-purpose AI tools can help with research on general topics, “they have important limitations that could lead to inaccurate or inappropriate suggestions” and recommended consumers always verify AI-sourced financial claims against trusted, licensed sources before acting.
That warning addresses consumer behaviour but stops short of platform liability. Under section 911A of the Corporations Act 2001 (Cth) and ASIC’s Information Sheet INFO 282, carrying on a financial services business in Australia without an AFSL is a criminal offence under sections 911A and 1311(1) of the Act, with responsibility resting on the unlicensed entity. The harder regulatory question – whether an AI platform that retrieves and presents content from unlicensed sources bears any liability for that output – has not been resolved.
ASIC’s Key Issues Outlook 2026, published January 27, 2026, frames the broader tension directly. The regulator identifies “advanced technology harming consumers (including agentic AI)” as a standalone 2026 priority, noting “variable maturity in how businesses manage AI governance risks.” On the licensing perimeter, ASIC states that “rapid innovation by or for people unfamiliar with financial services – particularly in digital assets and fintechs – continues to create risks including with unlicensed advice, misleading conduct, and the exploitation of unclear regulatory boundaries,” adding that “some entities will actively seek to remain outside regulation, contributing to perceived regulatory uncertainty.”
The Australian Treasury’s October 2025 Review of AI and the Australian Consumer Law found the existing framework “principles-based and technology-agnostic” and fit for purpose but placed the compliance onus on deploying businesses: the review found that “the onus is on all businesses to ensure the technologies they use are fit-for-purpose and must assess the risk that their systems or processes may mislead a consumer.” Whether AI platforms retrieving third-party content from unlicensed operators fall within that obligation is unaddressed. An international precedent sharpens the question. Somantra’s report notes a May 2026 ruling by the Regional Court of Munich, which found that Google’s AI Overviews constitute that company’s own editorial content rather than neutral pointers to third-party sources – a classification that, if applied under Australian consumer law, would shift liability toward the platform.
Research commissioned by the Council of Australian Life Insurers (CALI) and published in March 2026 found that three in five Australians would trust financial advice from AI tools such as ChatGPT. CALI CEO Christine Cupitt identified the risk directly: “Without the right kind of advice from the right people, Australians are at greater risk of falling victim to scams and dodgy providers.” The scam environment reinforces the stakes. ASIC’s April 2026 media release reported that between January and December 2025, the regulator coordinated the removal of 11,964 phishing and investment scam websites – a 90% increase on the prior period. Australians lost $2.18 billion to scams in 2025, with investment scams accounting for $837.7 million.
Arun Prasad, founder of Somantra, said the structural shift in how consumers receive insurance advice is the central issue for the industry. “AI search has compressed the insurance advice funnel from a marketplace of sources into a monologue. Consumers can no longer see whether a recommendation came from a licensed insurer, a comparison site’s methodology, or a content farm built to exploit how these systems retrieve information. For a category as consequential as insurance, that’s a visibility problem for brands and a trust problem for consumers,” Prasad said.
The Somantra report includes a Generative Engine Optimisation (GEO) playbook for Australian insurance brands and brokers, covering comparison-site data quality, structured content strategy, and monitoring frameworks for tracking AI-generated representations The full report, including domain-level data and methodology, is available on Somantra’s website.
