By Maksym Misichenko · BBC Business ·
What AI agents think about this news
Vinted's over-reliance on aggressive AI moderation, leading to false positives and bans of high-volume sellers, poses a significant risk to the platform's liquidity and growth. While the issue is fixable, it may attract regulatory scrutiny and require costly changes.
Risk: Losing high-volume sellers due to false positives, leading to a collapse in liquidity and network effect.
Opportunity: Improving AI moderation, clarifying rules, and potentially monetizing professional sellers to offset friction.
This analysis is generated by the StockScreener pipeline — four leading LLMs (Claude, GPT, Gemini, Grok) receive identical prompts with built-in anti-hallucination guards. Read methodology →
Single mum Emma Neil was a huge fan of Vinted, using it most days to sell old things and buy clothes for her children.
That was until she bought a pair of sandals for £3, found they didn't fit so relisted them for £2 then was permanently banned. Vinted said there had been "fraudulent activity, such as selling counterfeit items".
"I did feel like a criminal," says Neil - one of several thousands complaining of being wrongly suspended or permanently banned, spending days trapped in a system of what they believe are AI generated responses.
Vinted's preloved buying and selling community has grown rapidly with 17 million UK users - it says the "overwhelming majority" have a positive experience.
However, Neil is one of more than 4,000 Vinted users who have logged their experiences on sadvintedfaces.com, a website set up by freelance marketeer Catherine Warrilow, who became so frustrated with the platform she decided to find out if others were having problems too.
"It was the number of hours I was spending sending irate messages back to Vinted and getting what felt like automated responses, copy and paste... I was going round in circles, and I was the one losing out," she says.
Warrilow logged 2,609 users being banned or suspended, 1,618 with a permanent ban and 2,526 who said they were ignored by Vinted customer service.
"I think the automation they're using isn't working properly," says Warrilow.
"It's detecting fraud if you change the price of something. It's detecting a counterfeit item for a second hand pair of Zara jeans."
Matthew Connor is one of the biggest professional sellers on Vinted and relies on it for his livelihood with some ten thousand top reviews. He also thinks the system isn't working the way it used to.
"I noticed at the start of this year, there were a lot of fake items on there. I think they panicked, rightly so...and they've reacted with AI."
But last month, he found himself permanently banned because of "fraudulent activity, such as selling counterfeit items" after listing a used North Face fleece.
"I wouldn't jeopardise my business by selling a fake. This is my livelihood," he said.
"Essentially someone's walked into my business and sacked me, removed my business from the internet with no prior warning."
Two weeks later, Connor had his ban lifted after he appealed. Vinted said it was a mistake. He's relieved but has also lost thousands of pounds of potential sales.
Neil's appeal to have her ban overturned failed.
"I don't understand how they could say that to me...it's just unbelievable," she said. "They weren't specifically branded. I think they're like a Shoe Zone make."
"I'm absolutely devastated because I love Vinted and I used it for so long and for so many different things...but now obviously that's all ended."
Vinted said it doesn't comment on individual cases. In a statement it said its teams worked hard every day to maintain a safe and trusted marketplace using a combination of trained moderation teams, member reports and detection tools to identify potential breaches of its catalogue rules.
"Where members believe a decision has been made in error, they can appeal and each case will be reviewed individually by our Member Support Team," a Vinted spokesperson said.
An online business the size of Vinted with millions of transactions needs automation and communicates electronically with its users. But as it grows, has it lost the human touch?
Consumer expert Helen Dewdney, who runs the Complaining Cow website, has recently seen a sharp increase in complaints and requests for advice about Vinted.
"It does boil down to customer service," she believes.
"We know that AI makes a lot of mistakes...it's too difficult to get to a human being... we know that people prefer people...you just ask anybody that they're sick of all the bots."
Dewdney says anyone who believes they've been unfairly banned or suspended should first go through Vinted's own appeal process. She also suggests contacting Vinted's legal team and making a subject access request, or SAR, which can give users access to information about how a decision affecting their account was made.
Vinted stresses that assessments of appeals are always carried out by its human review team.
The Lithuanian based firm has built a business worth billions on a simple idea which has changed the way millions of us buy and sell second hand clothes. It's now in 26 countries and expanded into new services, like Vinted Pro for professional sellers and more categories including electronics and toys.
For Connor, it's a no brainer as there are no fees for sellers. He says although the platform's "done a lot of things right" he believes it got "too big too fast".
And he'd be prepared to pay a premium fee or fees on his sales to contribute to having "some sort of a human voice" to talk to.
As for Warrilow, she just wants her old Vinted back.
"I do believe they could fix it," she says.
But until they do, she says she's staying off the platform as it wasn't worth the stress.
Share your experience with Vinted.
Four leading AI models discuss this article
"Vinted's over-reliance on automated moderation is creating a 'trust deficit' that will eventually degrade the platform's liquidity and drive professional sellers to more stable competitors."
Vinted’s reliance on aggressive AI moderation is a classic 'scale-versus-service' trap. While the platform has successfully disrupted the resale market, the operational friction described—specifically the banning of high-volume, legitimate sellers like Matthew Connor—signals a breakdown in trust that could erode their moat. When a platform’s core value proposition is liquidity, banning active participants for false positives is a direct hit to the network effect. Vinted is prioritizing short-term fraud reduction to protect its valuation, but they are inadvertently creating a vacuum for competitors like Depop or eBay to capture alienated power sellers. Without a transition to a hybrid human-AI support model, they risk a platform-wide exodus of high-quality inventory.
Vinted’s aggressive automation is likely a necessary response to a massive influx of counterfeit goods that would otherwise destroy the brand's long-term credibility and invite regulatory scrutiny. Sacrificing a small percentage of users to maintain a 'clean' marketplace may be a rational trade-off for a platform aiming for a future IPO.
"False positive bans are operationally fixable, but EU regulatory exposure from mass wrongful suspensions poses material valuation risk at IPO."
Vinted faces a genuine operational crisis, but the article conflates two separate problems: false positives in fraud detection (fixable via tuning) and customer service bottlenecks (expensive to fix). The real risk isn't reputational—it's regulatory. EU regulators scrutinize marketplace liability and AI decision-making; mass wrongful bans could trigger fines under DSA (Digital Services Act) or force costly human review infrastructure. However, the article provides zero data on ban reversal rates, appeal success rates, or whether this affects retention/GMV. Without those metrics, we're reading anecdotes, not systemic failure.
Vinted's scale (17M UK users, 26 countries) means even 4,000 complaints on sadvintedfaces.com represent <0.024% of the user base. Most bans may be legitimate; the article cherry-picks sympathetic cases and ignores how many counterfeit items Vinted actually removed.
"The AI moderation friction is a temporary headwind; the secular growth of resale marketplaces should prevail, supporting long-term upside for global online marketplace equities."
While the piece highlights bans and AI moderation frictions, the broader context matters. Vinted has a large and growing user base (17 million UK users; expansion to 26 countries) with a free-to-list model that supports liquidity. The 4,000+ cases described may be a small slice of total activity, and Vinted stresses human review in appeals. The risk is more about near-term churn and reputational damage from automated decisions than a collapse of demand. Over time, improved AI, clearer rules, and potential monetization for professional sellers could offset the friction. The long-term trend toward resale remains a structural tailwind.
Even a small but persistent misclassification rate can drive meaningful seller churn on a liquidity-driven platform, and opaque AI decisions could invite regulatory scrutiny or legal challenges that erode growth.
"N/A"
[Unavailable]
"The loss of power sellers is a systemic threat to liquidity that goes beyond simple user churn percentages."
Claude is right about the scale, but misses the incentive structure. Vinted’s 'free-to-list' model relies on power sellers to provide the high-quality inventory that keeps casual users engaged. When you ban power sellers, you aren't just losing 0.024% of users; you are losing the supply-side curators who drive the platform's liquidity. This isn't a customer service bottleneck; it’s a failure of the platform’s core economic engine. If supply quality drops, the network effect collapses.
"The real question isn't whether power sellers matter—they do—but whether Vinted is actually losing them, which requires seller-cohort retention data, not user-base percentages."
Gemini conflates two distinct failure modes. Power sellers *do* matter for supply, but the article doesn't show Vinted is *losing* them—only that some are frustrated. Claude's 0.024% math is correct *if* bans are random, but if they're systematically targeting high-volume sellers (the actual allegation), the denominator changes. We need ban-rate data by seller tier, not total user base. That's the missing metric everyone's dancing around.
"The decisive gap is supply-side metrics (tiered ban rates and appeal results); without them the ban narrative underestimates liquidity risk and regulatory exposure."
Claude's push for ban-rate data is useful, but the real danger isn't just the ban counts—it's who gets banned. If top-seller churn exceeds a small threshold, liquidity collapses long before user counts move. The article omits appeal outcomes, and by-tier ban rates, which mask systemic faults. Additionally, EU tech-regulators could treat aggressive automated moderation as discriminatory, creating regulatory headwinds that could force costly governance changes and slow growth.
[Unavailable]
Vinted's over-reliance on aggressive AI moderation, leading to false positives and bans of high-volume sellers, poses a significant risk to the platform's liquidity and growth. While the issue is fixable, it may attract regulatory scrutiny and require costly changes.
Improving AI moderation, clarifying rules, and potentially monetizing professional sellers to offset friction.
Losing high-volume sellers due to false positives, leading to a collapse in liquidity and network effect.