Just in: What Australia's AI Trust Gap Means for Brands Selling Through AI Agents
By Malin Gaertig · September 9, 2026
Every week for months, the agentic commerce story has been about infrastructure — a new protocol here, a new payment rail there. This week the story flipped. New data out of Australia, alongside fresh moves from Amazon, Kohl's and one of the biggest landlords in America, made it clear that the technology is no longer the constraint. Consumer and customer trust is. If you sell anything through a channel an AI agent might soon touch, that shift changes what you should be building next, and it changes it now, not in some hypothetical future state.
- —Adyen's Australia data shows the trust gap clearly: 64 per cent of shoppers use AI to browse, but only 26 per cent will let it complete a purchase and 5 per cent will let it buy independently — winning discovery is not the same fight as winning the transaction.
- —Amazon's new Alexa scam-check tool signals the next phase is trust infrastructure — identity checks, verified messaging, fraud scoring — and verified communication channels are becoming table stakes as impersonation scales with agentic shopping.
- —Kohl's credited a Gemini-powered AI shopping assistant with measurable conversion and revenue-per-visit gains in an earnings call — use AI where shoppers already trust it (discovery, recommendation) and leave the final purchase with the human for now.
- —Simon's Media Network shows anyone holding first-party behavioural data can become a media business — expect new retail media entrants before most brands finish maturing their current strategy.
- —Audit where your brand asks a customer to trust an AI with money, not just a recommendation, and treat that moment as its own design and investment priority this quarter.
Every week for months, the agentic commerce story has been about infrastructure — a new protocol here, a new payment rail there. This week the story flipped. New data out of Australia, alongside fresh moves from Amazon, Kohl's and one of the biggest landlords in America, made it clear that the technology is no longer the constraint. Consumer and customer trust is. If you sell anything through a channel an AI agent might soon touch, that shift changes what you should be building next, and it changes it now, not in some hypothetical future state.
Why won't Australians let AI complete a purchase?
Adyen's 2026 Australia Retail Report landed this week with the sharpest local read yet on agentic commerce, and the numbers are worth sitting with. Every retailer surveyed is using AI in some form, and 95 per cent of enterprise retailers say they are already familiar with agentic commerce, where an AI agent discovers, compares and eventually buys on a shopper's behalf ([Inside Retail Australia](https://insideretail.com.au/digital/agentic-commerce-is-moving-from-concept-to-reality-202609)). Shoppers are moving too, with 64 per cent already using AI to browse or compare products.
Then the number that matters: only 26 per cent are comfortable letting AI complete a purchase, and just 5 per cent are fully comfortable letting it buy independently. Payments are where the hesitation concentrates — 62 per cent say a payment error damages how they see a retailer, and 91 per cent would take a slower, more secure checkout over a faster one.
For brands, this is the real headline. Winning discovery is not the same fight as winning the transaction, and most organisations are still resourcing only the first one.
Does agentic commerce need trust infrastructure more than smarter agents?
Amazon's answer this week was yes. Alongside a minor feature letting Alexa track restocks and price drops, Amazon quietly added something more consequential: a tool that lets shoppers ask Alexa whether a text, email or call claiming to be from Amazon is genuine ([Digital Commerce 360](https://www.digitalcommerce360.com/2026/09/04/amazon-alexa-for-shopping-update-me-when-fraud-prevention/)). Around 360,000 customers a year already contact Amazon support with exactly that question, and every check submitted now feeds Amazon's own fraud-enforcement team.
It is not a glamorous launch, and that is precisely the point. The most plausible next phase of agentic commerce is not more autonomous buying, it is the unglamorous trust layer underneath — identity checks, verified messaging, fraud scoring — that decides whether a nervous shopper hands an AI their card details at all. Brands should expect scam and impersonation attempts to scale in step with agentic shopping, and verified communication channels are quickly becoming table stakes rather than a differentiator.
Can an AI shopping assistant move the needle on real revenue?
Kohl's gave one of the clearer answers yet. In a Q2 earnings call where overall sales stayed soft, the retailer's digital business grew, and executives credited a Gemini-powered AI shopping assistant, rolled out for back-to-school after an earlier Mother's Day trial, with driving stronger conversion and higher revenue per visit among the shoppers who use it ([Digital Commerce 360](https://www.digitalcommerce360.com/2026/09/01/kohls-digital-sales-ai-shopping-conversion-q2-fy26/)). Adoption is still small, but it is one of the first US retail earnings calls where an AI assistant gets credit for a specific, measurable outcome rather than a vague innovation mention.
Read against the Adyen data, the playbook gets clearer: use AI to sharpen the parts of the journey shoppers already trust it with, discovery and recommendation, and leave the final purchase decision with the human for now. That is a more defensible near-term strategy than chasing full autonomy before the trust exists to support it.
Who else is about to become a media company?
Simon, the largest mall operator in the US, launched Simon Media Network this month, selling advertisers a closed-loop view across more than 200 physical destinations, its own marketplace, its loyalty programme and its digital channels ([Digiday](https://digiday.com/media/location-data-fuels-mall-giant-simons-new-pitch-to-advertisers/)). It is a different shape of retail media network, built on an ecosystem of shopping, dining and entertainment behaviour rather than a single retailer's transaction data.
The signal here is broader than one landlord. Anyone sitting on first-party behavioural data, payment networks and loyalty schemes included, is a candidate to become a media business next. The most likely path is that several of them move before most brands have finished maturing their existing retail media strategy, which quietly changes who you end up negotiating with for reach.
What this means for the year ahead
Put these four stories side by side and a single pattern emerges: the agents are ready to shop, and the businesses building for them are moving fast, but the people on the other end of the transaction are not ready to hand over full control, and may not be for some time yet. That gap will not close through better models. It will close through better payments experiences, clearer verification, and interfaces that make an AI agent's actions feel accountable rather than opaque.
The practical takeaway is simple enough to act on this quarter: audit where your brand currently asks a customer to trust an AI with money, not just with a recommendation, and treat that single moment as its own design and investment priority, separate from the shopping experience around it. The brands that get comfortable with that distinction now will be the ones agentic commerce actually rewards, whichever protocol ends up winning the standards fight underneath it.