Un abonnement mensuel de 10$, 20$ ou 35$ peut faire une grande différence.

Vous apercevez quelque chose que nous devrions vérifier ? Écrivez-nous.

Analysis: Nonexistent customers and AI ragebait promos attempt to lure conservatives into buying overpriced US patriotic merch

Social media, AI and Shopify are exploi­ted to accele­rate mass e‑commerce manipu­la­tions aiming to extract money, as consu­mer protec­tion resear­chers long warned would happen.

A young blond woman comes to a local market to sell handmade necklaces but immedia­tely gets punched in the face by a stran­ger she identi­fies as a “Democrat” (archi­ved here) – a detail sugges­ting the attack was linked to the Ameri­can flag displayed on her merchan­dise. This is the story­line repea­ted across multiple videos posted by the @rubycrafts5 account (archi­ved here) on TikTok. 

But is it true? 

VérifRa­dar decons­truc­ted the decep­tion piece by piece and found that none of this happe­ned in real life. The ragebait videos, designed to trick conser­va­tive audiences into buying patrio­tic pendants, were AI-genera­ted. Some also targe­ted Chris­tians and the U.S. army, deploying a variety of tactics with notable resem­blances. 

Are platforms doing enough?

Direc­tor of the Indiana Univer­sity Obser­va­tory on Social Media Filippo Menczer (archi­ved here), on being asked if platforms are doing enough to disrupt manipu­la­tion, said: 

“Platforms have been slow and reluc­tant to adopt policies to deal with AI-enabled manipu­la­tion and abuse, whether it’s nudifiers or infor­ma­tion opera­tions or fraud. For example, we uncove­red a large AI-powered malicious botnet perfor­ming crypto fraud on Twitter/X and it was removed only after we publi­shed the results.” 

He conti­nued:

“Worse, they removed trust and safety teams, aboli­shed fact-checking programs, and incen­ti­vi­zed AI-genera­ted engaging content (slop) through revenue-sharing. Platforms have also resis­ted regula­tion from outside the U.S., such as the EU. So no, they are defini­tely not doing enough to curb the harms of AI manipu­la­tion. We have daily reports of harmful conse­quences, from fake doctors giving unvet­ted (or dange­rous) medical advice to fake images and videos for election manipu­la­tion.”

A case analysis

This is what one of such posts (archi­ved here) that went live on July 22, 2026, looked like on TikTok at the time of writing:

(Image source: post by @rubycrafts5 on TikTok.)

Sighten­gine, an AI detec­tion tool, said that the clip was 89% likely to have been genera­ted by AI: 

(Image source: Sighten­gine.) 

Hive Modera­tion, another detec­tion tool, ruled that the odds of the clip having been made by AI were 99%:

(Image source: Hive Modera­tion.)

When VerifRa­dar tested the video’s thumb­nail image with ZeroGPT, it retur­ned a similar result. Accor­ding to that tool, the thumb­nail was “96% FAKE”:

(Image source: ZeroGPT.)

In the video, the same necklace showed a different number of stars on the U.S. flag normally featu­ring 50 stars to represent each of the 50 states (archi­ved here). At first, the pendant displayed 45 stars, then, near the end of the video, the total went down to 41 stars.

(Image source: post by @rubycrafts5 on TikTok.)

(Image source: post by @rubycrafts5 on TikTok.)

The shop said nothing about offering a custo­mi­zable number of states on the Ameri­can natio­nal flag. Yet, at the begin­ning of the video, the necklaces showed no stars at all:

(Image source: post by @rubycrafts5 on TikTok.)

VérifRa­dar found another account (archi­ved here) on TikTok that direc­ted to the same store, but it showed a different influen­cer. Yet, that person was also created by AI: 

(Image source: Sighten­gine.)

Several comments in one video that gained a lot of traction from that account (archi­ved here) demons­trate indica­tions that corres­pond to coordi­na­ted inauthen­tic behavior, with a number of users asking to have the necklaces shipped to Israel, or “sending love from Tel Aviv”. 

Those accounts’ bio don’t show them living in that country in some cases, inclu­ding one from the Nether­lands, one from Canada, one more from Mexico and one displaying support for Pales­tine. That suggests many of the accounts may have been hijacked and could be used by a bot farm to trick the algorithm and inflate engage­ment as a result.

The pendant-selling website (archi­ved here) adver­ti­sed by the account whose videos are the focus of this article was regis­te­red on July 24, 2026, accor­ding to the WhoIs lookup tool (archi­ved here):

(Image source: WhoIs.com.)

However, one review marked as “verified” claimed that its author was able to order a patrio­tic necklace from the page 20 days before its regis­tra­tion: 

(Image source: myusacrafts.com.)

ZeroGPT found that the image of the suppo­sedly real custo­mer was 97% likely to have been created by AI:

(Image source: ZeroGPT.com.)

Accor­ding to ImageW­his­pe­rer, the odds of that person’s “photo” being a product of genera­tive AI were 100%. The tool further pointed to a speci­fic model used to make the picture: Gemini 3.

(Image source: ImageW­his­pe­rer.)

Gemini, in turn, also confir­med that. It said: 

“Yes, this image was genera­ted or edited using Google AI.”

(Image source: Gemini.)

While the website showed that it has 93 reviews avera­ging a rating of 4.8 stars, clicking on the “show more reviews” button genera­ted a loop of the same handful of reviews.

As of August 4, 2026, the website displayed “4th of July” 50% off sales with a timer that resets every­day to give a sense of urgency for custo­mers to buy the products. It also offered to tip and to get “mystery packages” for an extra price or at no extra charge if more necklaces were purcha­sed.

AI is accele­ra­ting the scope and scale of the manipu­la­tions that aim to extract money from unsus­pec­ting victims browsing so-called “hand-crafted” products, Filippo Menczer said:

“Research from our Obser­va­tory on Social Media has been raising aware­ness about the possible harms of AI-enabled manipu­la­tion…” 

He went on:

“…what makes this kind of fraud concer­ning is that AI is used to generate engaging fake content (ragebait videos, reviews and comments) targe­ting a parti­cu­lar commu­nity — MAGA people in the examples you shared with me. The same AI tools can be used to quickly and cheaply generate this kind of manipu­la­tive content targe­ting any commu­nity to steal their money or finan­cial infor­ma­tion.”

Although the online store reads “Support a small Business” and the AI video claims the necklaces are handmade, a Google Lens search of the necklaces provi­ded results of similar widely available products going for far less than the price it lists. The same necklaces can be bought from Temu for three Canadian dollars (archi­ved here) or for two Ameri­can dollars (archi­ved here) on Alibaba. They are also available for a fraction of the cost on Etsy (archi­ved here) at USD 8,40 from a seller shipping from the U.S.

VérifRa­dar attemp­ted to contact the phone number provi­ded in the “Contact Infor­ma­tion” section of the website three times. The call went through, but no real human picked up the phone. 

Sending credit card infor­ma­tion to unknown merchants, especially if invol­ved in posting fake videos, creates risks of fraud.

Phone number reused

VérifRa­dar found multiple online stores displaying the same phone number, but each time it was attri­bu­ted to different names. A Google search of the phone number found on myusacrafts.com shows multiple Shopify stores indica­ting the same number as their contact infor­ma­tion, also displaying similar patterns of manipu­la­tion:

(Image source: Google.)

More stores targeting Christians and the U.S. army service members

The case of the necklace-selling enter­prise was one of many examples illus­tra­ting a much broader pattern on social media. 

Previously, in mid-summer 2026, another fact-checking organi­za­tion, Lead Stories, wrote about two other fictio­nal charac­ters, “Lily” and “Taylor” claiming to have been beaten while selling Jesus hoodies. Just like the necklace-selling woman, they were created by algorithms.

The AI-genera­ted artisans typically follow the same generic appel­la­tions, with names that sound familiar to Western audiences along with words such as “crafts”, “creates” or other synonyms and the fake hand-crafters represent presu­ma­bly white people. Moreo­ver, they all offer sham reset­ting sales hovering around 50%.

By searching for the keywords that inclu­ded common Western first names along with “crafts” on TikTok, VérifRa­dar quickly found a store (archi­ved here) from an account (archi­ved here) that sells “God is good” hats, again with an unori­gi­nal design that can be found for a much lower price tag on sites such as Etsy (archi­ved here). 

Origi­nally, that account on TikTok direc­ted to thewesterncrafts.shop for handmade leather hairbrushes, but now it redirects to another store that sells no hairbrushes. Although the store’s domain name is joshcrafts.com (archi­ved here), its descrip­tion reads “Crafts By Sophia”. That was regis­te­red through Tucows and Contact Privacy Inc., per the WhoIs record, in July 2026 – just like myusacrafts.com. The hats sold there can be found for around ten dollars USD from Walmart (archi­ved here) or Amazon (archi­ved here).

VérifRa­dar found that at least one more – cap-adver­ti­sing – website (archi­ved here) showed the same patterns with the fake “4th of July” sale timers as the ones selling patrio­tic pendants. Just like the necklaces, the patrio­tic caps could be found with Google Lens for cheaper. Inter­es­tin­gly, that website was also regis­te­red in July 2026, as WhoIs records (archi­ved here) revea­led: 

(Image source: WhoIs.)

Other e‑commerce manipu­la­tions (archi­ved here and here) associa­ted with AI-genera­ted brand ambas­sa­dors (archi­ved here and here) on TikTok targe­ted the U.S. Army. VérifRa­dar contac­ted a Penta­gon Duty Officer for comment. This article will be updated as appro­priate when a response is recei­ved.

More similarities: Canada ties and the same site-building app

Many of the sites revie­wed by VérifRa­dar shared other features as well. 

For example, those stores were all built through Shopify, the websites’ code shows: 

(Images’ source: HTML source of three stores revie­wed in this analy­sis.)

Another simila­rity was that at least four of those sites were regis­te­red during the same time window (in late June and throu­ghout July) through Contact Privacy Inc., a privacy safety company based in Toronto. For a relati­vely small extra fee, it allows owners of a webpage to keep their WhoIs infor­ma­tion private. 

The repea­ted domain regis­trar was also a Toronto-based company, Tucows Domains.

VérifRa­dar provi­ded its findings sugges­ting inauthen­tic behavior to the domain regis­trar of the merch websites and the WhoIs privacy company, but did not hear back days later. TikTok and Shopify were also notified. If we receive a response, this article will be updated as appro­priate.