Alexa for Shopping can now tell you whether that suspicious Amazon message sitting in your inbox is genuine or a fraud. Amazon announced the new capability on 2 September 2026: ask its shopping assistant about any email, text or phone call that claims to come from Amazon, and the AI will check it against the company’s own records and give you a verdict within seconds. For a brand that ranks among the most impersonated on the planet, it is a quietly significant move.

The scale of the problem explains the timing. Roughly 360,000 customers contact Amazon customer service every single year just to ask whether a message they received is real. Meanwhile, the US Federal Trade Commission says Americans reported losing a record 3.5 billion dollars to imposter scams in 2025, with losses nearly tripling since 2020. Fraudsters lean on generative AI to write cleaner phishing lures, and the gap between a real notification and a fake one keeps narrowing.

This article breaks down how the new verification feature inside Alexa for Shopping works, the fraud statistics behind it, how to use it today, and what it signals for any organisation thinking about its own cybersecurity posture and AI strategy.

What Amazon Announced

alexa for shopping scam message verification b solid envelope closed flap

Amazon unveiled the scam-checking capability through its About Amazon newsroom, and TechCrunch’s Sarah Perez covered the launch the same morning under the headline “PSA: Amazon’s shopping AI can now tell you if that message is a scam.” The feature lives inside Alexa for Shopping, the agentic assistant Amazon rolled out earlier this year.

A scam checker inside the shopping assistant

Rather than shipping a standalone security product, Amazon folded message verification into the assistant customers already use to compare products and track orders. Describe the message you received to Alexa for Shopping, and it cross-references the communication against the billions of messages Amazon teams have sent globally. Amazon says the system can “definitively” confirm whether a communication originated from its own systems.

Who announced it

Scott Knapp, Amazon’s vice president of Worldwide Buyer Risk Prevention, framed the problem bluntly: “Scammers thrive when customers can’t easily verify whether a message is real.” His team built the tool to close exactly that verification gap, turning a slow customer-service enquiry into an instant AI-powered lookup.

Why it matters now

Impersonation fraud has become the single largest category of consumer fraud reported in the United States. Nearly one in three fraud reports filed with the FTC in 2025 concerned an imposter of some kind. When a checkout-scale platform gives every customer a free verification tool, it meaningfully raises the cost of running a fake-Amazon operation.

How Alexa for Shopping Verifies a Suspicious Message

alexa for shopping scam message verification c padlock shackle shut flush

Under the hood, the feature is a matching engine. Alexa for Shopping compares what you describe against Amazon’s canonical record of outbound communications, then reasons about whether your message could plausibly have come from the company.

The signals it checks

The AI analyses sender information, message content, timing, formatting, delivery method and message metadata. A real Amazon delivery notification follows predictable patterns: it comes from known addresses, references genuine orders, and lands at times that correlate with actual fulfilment events. A fake usually gets at least one of those signals wrong, even when the wording looks flawless.

Billions of messages as ground truth

Because Amazon operates its own email, SMS and phone infrastructure, it holds something no third-party scam filter has: a complete ledger of everything it actually sent. Alexa for Shopping checks your message against that ledger. Amazon adds that the system improves as customers report more suspicious messages, so every fraud attempt fed into it sharpens detection for everyone else.

Three verdicts in seconds

You provide the details — where the message came from, when it arrived, what it says — and receive one of three responses, typically within seconds.

VerdictWhat it meansWhat you get next
Confirmed from AmazonThe message matches Amazon’s outbound recordsReassurance plus tailored security tips
Not from AmazonNo match — the message is an impersonationActionable next steps to protect your account
Unable to verifyNot enough detail to reach a confident answerGuidance on what to resubmit for a clear verdict

That three-way outcome matters. A tool that only ever said “looks fine” or “looks risky” would train users to ignore it; an explicit “unable to verify” keeps the honest uncertainty visible.

The Scale of the Amazon Impersonation Problem

alexa for shopping scam message verification d blank speech bubble tail

The verification feature did not appear in a vacuum. Amazon sits near the top of every brand-impersonation league table, and the losses behind those tables keep climbing.

360,000 questions a year

Amazon’s own figure is striking: about 360,000 customers a year contact customer service purely to ask whether a message is authentic. Each of those contacts represents a person who paused before clicking — and thousands more presumably never pause at all. Automating that check removes the friction that stops people verifying in the first place.

What the FTC numbers show

The FTC’s 2025 data paints the wider picture. Reported imposter-scam losses reached 3.5 billion dollars, the highest figure on record. Within that total, business impersonators accounted for nearly 1 billion dollars in reported losses, while government impersonators took about 920 million dollars.

The chart below shows how those reported 2025 losses break down, using the FTC’s own figures.

Reported US imposter-scam losses, 2025 (FTC)
All imposter scams $3.5bn
Business impersonators ~$1.0bn
Government impersonators ~$0.92bn

Amazon among the most impersonated brands

In earlier FTC brand-level data, consumers filed about 34,000 reports of scammers impersonating Amazon in a single year — second only to the roughly 52,000 reports naming Best Buy and its Geek Squad brand. The comparison below uses those FTC report counts.

Brand impersonation reports to the FTC (2023)
Best Buy / Geek Squad 52,000
Amazon 34,000

Older adults are hit hardest

The FTC has also documented a more than four-fold increase in reports from older adults who lost tens or even hundreds of thousands of dollars to impersonation scammers. For that group, a spoken conversation with an assistant that answers “was that call really from Amazon?” may be far more usable than headers, hover-checks and reporting portals.

The Scams Alexa for Shopping Is Built to Catch

alexa for shopping scam message verification e card index box five cards

Amazon lists five recurring impersonation patterns that the verification feature targets. Anyone who shops online will recognise most of them, and Alexa for Shopping now gives each one a fast, authoritative counter.

Scam patternThe hookHow verification defuses it
Fake order confirmations“You bought a $900 laptop — call us to cancel”No matching order or message in Amazon’s records
Prime membership renewalsUrgent payment-failure or renewal alertsConfirms whether Amazon actually sent a Prime notice
Account suspension warnings“Verify your details or lose your account”Flags the message as not originating from Amazon
Package delivery notificationsFake failed-delivery texts with malicious linksChecks timing against real fulfilment events
Job offer scamsBogus Amazon recruitment with upfront feesConfirms Amazon recruiting never sent the offer

Why these keep working

All five patterns exploit the same reflex: urgency plus a plausible brand. A shopper with several live orders cannot instantly recall whether a delivery text is expected, and scammers deliberately time their blasts around peak shopping seasons. The counter is not more suspicion — it is cheap, instant verification, which is precisely what Alexa for Shopping now supplies.

The phishing economics behind them

Impersonation is popular because it scales. One convincing template can hit millions of inboxes, and phishing kits sold in underground markets now bundle brand-perfect Amazon assets. A phishing campaign only needs a fraction of a percent of recipients to click to be profitable, which is why defenders focus on cutting response rates rather than stopping every send.

How to Use Alexa for Shopping as a Scam Checker

alexa for shopping scam message verification f round compass blank dial

Using the feature requires no setup, no Echo device and no Prime subscription. If you can sign in to Amazon’s website or mobile app in the US, you can ask the assistant to vet a message right now.

What to ask

Amazon’s examples show the conversational shape of a check. You might ask: “Did Amazon send me a text about a delivery problem yesterday?” Or: “Is this email about my Prime membership from Amazon real?” Or even: “I got a call about a refund from someone saying they’re from Amazon. Was that real?” Provide where the message came from, when it arrived and what it says, and the verdict comes back in seconds.

Other verification channels

The AI check is the fastest route, but not the only one. Amazon maintains a dedicated address, [email protected], where customers can forward suspicious emails, plus a verification form on its customer-service pages. The table compares the options.

ChannelBest forSpeedEffort
Alexa for ShoppingAny message, text or call you can describeSecondsOne conversational question
[email protected]Forwarding a suspicious email intactSlower, asynchronousForward and wait
Customer-service formStructured reports and follow-upSlowestForm fields and navigation

Available in 28 languages

Amazon says verification support is available in 28 languages, which matters for a fraud category that deliberately targets people shopping in their second language. The launch covers US customers first, with expansion to additional regions planned.

From Rufus to Alexa for Shopping

The scam checker is also a story about where Amazon’s consumer AI is heading. To understand why verification landed here, it helps to know what Alexa for Shopping actually is.

A short history

Amazon launched Rufus, its first generative AI shopping chatbot, in 2024; the company says Rufus helped more than 300 million customers research, compare and buy products during 2025. In February 2025 Amazon shipped Alexa+, its upgraded assistant built on newer foundation models. Then, on 13 May 2026, Amazon retired the Rufus brand and launched Alexa for Shopping — a single assistant, powered by Alexa+, embedded directly in the search bar across mobile, desktop and Echo Show displays.

An agentic assistant, not just a chatbot

Alexa for Shopping goes beyond answering questions. It compares products, tracks prices, schedules recurring orders for essentials, and can even complete purchases on a user’s behalf — including, in some cases, beyond Amazon’s own marketplace. It is free for every signed-in customer, with no Echo device or Prime membership required. That agentic design philosophy mirrors what businesses are building with AI employees and autonomous agents across their own operations.

Why verification fits an agentic assistant

An assistant that can spend your money needs your trust, and trust is exactly what impersonation scams erode. By teaching Alexa for Shopping to police fake Amazon messages, the company is protecting the channel its agentic commerce strategy depends on. Every scam that lands makes customers warier of every legitimate notification — and warier of letting an AI act for them.

What It Means for Shoppers

For consumers, the practical takeaway is simple: stop guessing. Before you click a link, call a number back or “confirm” a payment, ask the assistant first.

Build the reflex

Security teams call this a verification habit, and habits need low friction to form. Asking Alexa for Shopping one question is easier than inspecting email headers, so the habit has a real chance of sticking — especially for the older adults the FTC identifies as the highest-loss group.

Keep the classic rules

The tool complements, rather than replaces, the basics. Amazon will never ask for payment over the phone, never request gift cards, and never ask you to install remote-access software. If a message pressures you to act in minutes, that urgency is itself the tell. A “not from Amazon” verdict should end the interaction immediately.

A checklist before you click

A sensible routine looks like this. First, do not act on the message itself — no links, no reply, no call-back. Second, open the Amazon app or website directly and ask Alexa for Shopping whether the communication is genuine, giving the source, arrival time and content. Third, follow the verdict: security tips if it is real, the recommended next steps if it is not. Finally, forward anything email-shaped to [email protected] so the detection system learns from it. The whole loop takes under a minute, which is the point — Alexa for Shopping only reduces fraud if checking is faster than falling for the scam.

What It Means for Businesses

The launch carries a lesson well beyond Amazon’s marketplace: brand impersonation is now a first-class product problem, and AI verification is becoming a standard counter.

Your brand will be spoofed too

Business impersonators extracted nearly 1 billion dollars from US consumers in 2025 alone. Any organisation that emails invoices, delivery updates or password resets is a candidate for spoofing. The defensive playbook Amazon just demonstrated — canonical message records plus an AI matching engine plus a customer-facing checker — is one that banks, carriers and retailers can expect to copy.

Verification as customer-facing infrastructure

Most companies treat fraud reporting as a cost centre buried three menus deep. Amazon turned it into a conversational feature with a measurable payoff: 360,000 annual support contacts that can now resolve themselves in seconds. Businesses exploring intelligent automation should note the pattern — the same AI plumbing that answers product questions can absorb high-volume trust-and-safety work.

Getting your own house in order

Before building an AI checker, most organisations need the fundamentals: authenticated email domains, consistent sender addresses and a security posture that makes genuine messages verifiable in the first place. That groundwork is exactly where a structured cybersecurity review pays off, and it is the unglamorous half of what makes Amazon’s system able to say “definitively” at all.

Limitations and Open Questions

The feature is genuinely useful, but it ships with boundaries worth understanding before you rely on it.

US-only at launch

Verification through Alexa for Shopping is currently launching for US customers. Shoppers elsewhere still depend on [email protected] and the reporting form until the promised regional expansion arrives — even though impersonation fraud is thoroughly global.

“Unable to verify” is a real outcome

The system needs enough detail to match a message against its records. A vague description of a phone call may come back unverifiable, with guidance to resubmit. That honesty is welcome, but it means the tool works best when you still have the message in front of you.

Scammers will adapt

Fraud is adversarial. As verification becomes routine, expect scams that impersonate the verifier itself — fake “Amazon security check” pages or calls claiming to be the scam checker. The safe pattern never changes: initiate the check yourself, inside Amazon’s own app or website, rather than trusting anything inbound. Amazon’s bet is that its feedback loop, fed by every reported fake, adapts faster than the attackers do.

What Alexa for Shopping does not do

The assistant verifies whether Amazon sent a message; it does not scan your inbox, block senders or judge messages claiming to come from other companies. A fake bank alert or a bogus courier text from another brand is outside its records, so Alexa for Shopping cannot rule on it. It also depends on your description being accurate — if you paraphrase a message loosely, the match may fail. Treat it as one strong tool in a wider habit of scepticism, not a complete spam filter, and keep your email provider’s own phishing protections switched on alongside it.

FAQ

Is the Alexa for Shopping scam checker free?

Yes. It is available to any customer signed in to Amazon’s website or mobile app in the US. There is no charge, no Echo device requirement and no Prime membership requirement.

What kinds of messages can Alexa for Shopping verify?

Emails, text messages and phone calls that claim to come from Amazon. You describe the source, timing and content, and the assistant checks the details against the billions of messages Amazon has actually sent, covering order confirmations, Prime notices, delivery alerts, account warnings and job offers.

How accurate is the verification?

Amazon says it can “definitively” confirm whether a communication originated from its systems, because it checks against its own outbound message records. When there is not enough information, the tool explicitly answers “unable to verify” rather than guessing.

What should I do if the verdict is “not from Amazon”?

Do not click links, call numbers or share any details from the message. Follow the next steps the assistant provides, report the message so the system learns from it, and delete it. If you already interacted with the scam, change your Amazon password and contact your bank promptly.

Does this replace Rufus?

Rufus was retired in May 2026 when Amazon launched Alexa for Shopping, which absorbed its product-research role and added agentic features. The scam-verification capability is new and did not exist in Rufus.

Where is Alexa for Shopping available?

Alexa for Shopping works for signed-in customers in the United States across the Amazon mobile app, the desktop website and Echo Show smart displays. The message-verification capability is part of the US launch, supports 28 languages, and Amazon says expansion to additional regions is planned. Customers outside the US can still forward suspicious emails to [email protected] in the meantime.

References