As Amazon tries to keep AI crawlers off its website, human customers are becoming collateral damage.
Since late last year, some Amazon shoppers have only been able to read a small number of customer reviews for each product. When this happens, the customer must file an email appeal to Amazon and wait several business days for a response.
An Amazon spokesperson confirmed that the company is trying to restrict unauthorized data scraping on its site, but it’s also mistakenly flagging legitimate customers as bots, with complaints piling up on Reddit and on Amazon’s forums. I’ve heard personally from a couple of affected users, one of whom has been a Prime subscriber since the program’s inception.
Amazon declined to answer questions about when it started restricting review access, how it verifies users’ requests to see more reviews, how many users have been affected, and what behaviors might result in a false positive, if any.
“We understand there may have been isolated instances where customers were unable to access reviews, and we’ve worked to ensure access,” a spokesperson tells Fast Company.
Even if most users never lose access to Amazon reviews, this kind of issue may soon become inescapable. As AI crawlers and agents get better at passing as human, and platforms keep trying to block them, proving that you’re not a robot is only going to get harder.
Blamed by customer service
Fred Hall is a longtime Prime subscriber who depends on Amazon’s customer reviews for product research. He searches through reviews for details on specific features, and he looks through reviews with low ratings to identify potential problems.
But when he went shopping on Amazon for a dehumidifier in mid-July, he found that Amazon would only let him see eight reviews for each product. An Amazon customer service chat agent informed him that he’d “violated Amazon’s Conditions of Use regarding data gathering or extraction of community-generated content, such as Customer Reviews.”
Hall calls this explanation “ridiculous.”
“Seems like the AI algorithm has been tuned a little too tight. I’ve been an Amazon Prime Member since Prime first became a thing,” he says. “I am not some data scraper trying to steal review data.”
While Amazon accepted Hall’s appeal a few days later, the company never explained what went wrong in the first place. He claims not to use any AI browsing or automation tools. The whole episode has him rethinking his Prime membership.
“If I hadn’t pushed in my chat with the live agent, I would have had no idea why my access was taken away in the first place, and even then the reasoning was pretty vague,” Hall says.
Similarly, Sean Kaufman found that he could only access eight reviews for any product, with no sorting or filtering options. Reached via Reddit after he aired his complaints on the site, Kaufman says he doesn’t use an ad blocker, a VPN, or any AI browsing tools. He has no idea why Amazon flagged his account, other than that he dives deeply into reviews from competing products while researching what to buy.
In Kaufman’s case, an Amazon chat agent offered a $10 promotional credit as “a token of apology” and promised to escalate the issue within 24 hours. But that never happened, and it took five days to get full review access again.
Kaufman says the lack of transparency from Amazon is what bothers him most, as he’s now left guessing what will happen if his shopping patterns “don’t meet algorithmic expectations of normalcy.”
“The thing that irritates me the most is that Amazon didn’t even bother to tell me my account had been restricted,” Kaufman continues. “Nothing was ever explained—before, during, or after—which has led to major confusion about the nature of this program.”
More AI bots, more hassles for humans
As AI agents and crawlers get better at mimicking human behavior, Amazon isn’t alone in inconveniencing humans as it tries to sort through who’s a real person online.
In April, Google started testing a new ReCaptcha system for curbing bot activity on websites. Instead of just making users solve a visual puzzle, the new system asks them to scan a QR code with their phones, while another new version asks users to perform a series of hand gestures. Google says those measures are specifically a response to AI agents that can autonomously browse and interact with the web—like the kind that Google itself is creating.
Meanwhile, Reddit has blocked nearly all search engines from accessing its site in an attempt to keep AI crawlers out. The lone exception is Google, which is paying to train AI on Reddit’s data. The Internet Archive, an invaluable source for referencing old versions of websites, has also been unable to archive stories from several news outlets that say they’re wary of AI scraping the content on their sites.
As for Amazon, in November it sued AI company Perplexity over an “agentic” shopping feature that browses Amazon on behalf of users. Amazon accused Perplexity of trying to make its AI agents pass as humans and of ignoring Amazon’s requests to turn the feature off. An Amazon spokesperson declined to comment on whether its clash with Perplexity has anything to do with its crackdown on review access. The company has yet to publish any documentation for users it has accused of scraping reviews.
Tech companies have promised a future in which visiting websites will be obsolete because AI will do it for you. Only now, they’re trying to block competitors from doing exactly that, and in the process they’re hindering the folks who actually want to visit their sites in the first place.
For Kaufman, the outlook is pretty bleak. “AI-driven decision-making is a scary prospect for humanity if this is any indication of how it’s going to go,” he says.
