The consumer regulator has put a name to something Australian retailers have been complaining about for two years. Ghost stores, online shops that pose as local businesses and are usually run from overseas, have stopped looking obviously fake. The reason, the ACCC says, is that generative AI has made a convincing shopfront cheap.
The warning is aimed at shoppers. The people it should worry most are the businesses being copied, because the damage lands on a brand that had no part in it and no easy way to stop it.
What the regulator actually said
An ACCC spokesperson put it plainly: the rise of AI tools is making it easier for scammers to create convincing fake websites, advertisements, images and content that appears professional and legitimate. AI can generate professional looking online stores, product images, customer reviews and marketing material, which makes it harder for consumers to tell a genuine business from a dodgy operator.
The line that matters is the next one. Consumers can no longer rely on appearance alone when deciding whether a business is legitimate, the regulator told SmartCompany. Scammers have always impersonated trusted brands. What changed is that AI lowered the cost and effort of doing it.
What a ghost store looks like
The pattern is consistent. A social media ad announces that a much loved local shop is closing down, with heavy discounts on the last stock. The site looks like a small Australian boutique, complete with a founder story, a Melbourne or Adelaide address and a wall of five star reviews. There is no shop. The goods, if they arrive, are low quality items drop shipped from overseas.
The ACCC has acted on this before. In July 2025 it issued public warning notices against four websites: everly-melbourne.com, willowandgrace-adelaide.com, sophie-claire.com and doublebayboutique.com. Deputy Chair Catriona Lowe said the regulator was warning Australians about those four specifically, which it alleged were not based in Melbourne, Adelaide or Double Bay, nor imminently closing down.
The scale behind that action is the part worth holding on to. From the start of 2025 the ACCC received more than 360 reports about roughly 60 online retailers, according to its media release. It also wrote to Meta and Shopify asking them to scrutinise ghost store operators.
Enforcement does move the needle, briefly. Around December 2025 roughly ten suspected ghost stores stopped operating, and the ACCC has acted against four more in the past year. Set against 60 retailers in the reports, that is a regulator working through a queue rather than clearing one, which is the shape of the problem when the cost of standing up the next site is close to nothing.
Why AI changed the economics
Faking a shopfront used to take real work. Someone had to write the product descriptions, invent the founder story, source photography that did not reverse image search back to a stock library, and seed enough plausible reviews to pass a glance. That effort was the filter. It kept the volume down and left tells behind.
Generative tools removed the filter. The copy, the imagery, the reviews and the ads now come out of the same afternoon, in a house style that reads like any other small brand. Nothing about this requires sophistication, which is why the volume has moved rather than the technique.

The business being copied carries the cost
Most coverage frames this as a consumer problem. For the impersonated business it is worse, because the complaints arrive at the real shop. Customers who paid a fake site turn up asking about a refund the actual owner never took, and the reviews they leave attach to the genuine brand.
There is no fast remedy either. Takedowns run through the platform hosting the ad and the platform hosting the store, which is why the ACCC wrote to Meta and Shopify rather than to the operators. A small retailer chasing that on its own is looking at weeks, and by then the site has moved.
Why Australian businesses should care
The practical exposure is not limited to retail. Any business with a recognisable name, a local story and photographs of real premises has everything an impersonator needs, and service businesses are no safer than shops. Trade names, accreditations and staff photos are all public.
Three things are worth doing before it happens rather than after. Search your own business name alongside terms like closing down and clearance every month, because that is where the ads surface first. Keep the Australian Business Register listing current so customers verifying you find a match. That record also holds your main business activity, so confirm its ANZSIC code still matches what you sell. And know in advance who inside the business owns a takedown request.
It is also worth being deliberate about how your brand reads to a machine. Consistent details across your site, your listings and your profiles are what makes a copy detectable, and that consistency is the same work that decides how you show up in AI generated answers. The two problems share a foundation.
Impersonation is the pattern to watch
This is not the first time this year that cheap generation has turned into impersonation at volume. In August we covered attackers posing as ClaudeBot to hunt for API keys, which is the same idea aimed at infrastructure rather than shoppers. The trust signals people rely on are the ones that got cheap to fake.
The uncomfortable conclusion is that looking legitimate has stopped being evidence of anything. Verification has to move to things that are hard to fabricate: registration records, independent reviews outside the seller’s own site, and a contact route that answers. For businesses planning where AI helps and where it creates exposure, that belongs in the same AI strategy conversation as everything else.
