QR code phishing declined sharply in the third quarter of 2026, according to Microsoft's latest email threat report, but the same report documents a single August campaign that used a QR code to reach more than 10,000 people. The takeaway for anyone who sends or scans codes is that a falling volume does not make a QR code safe to trust on sight. This article sets out the figures Microsoft published, walks through the campaign in plain language, and gives an honest account of what a dynamic QR platform can do about it. The short answer is that it cannot filter anyone's inbox, but it can help genuine senders stay recognisable and spot odd behaviour on their own codes.
Disclosure: we make D-QR, a dynamic QR code platform. This article is about attacks that happen in other people's email, which no QR tool, ours included, can block. We say where our product helps and where it does not.
What Microsoft reported about QR code phishing
On 5 November 2026, Microsoft Threat Intelligence published its email threat landscape report for the third quarter, covering July to September. The overall picture is of rising email threats alongside falling QR code activity. Microsoft says it detected roughly 7.3 billion email-based phishing threats across the quarter, with nearly 3 billion in September alone, a 41% rise on August.
QR code phishing moved the other way. Microsoft reports about 12.1 million QR code attacks during the quarter, down 67% from the second quarter. By September, monthly volume was 86% below its March peak. Much of that fall came from attachments, which had driven the earlier surge. PDFs were still the most common carrier, at 78% of QR code attacks in July and August and 61% in September, but the volume of PDFs with malicious codes fell by roughly 65% by the end of the quarter.
The shift inside the decline
One detail is worth noticing. Microsoft says QR codes placed directly in the body of an email grew from 5% of attacks in July to 12% in September, and that the volume of those more than tripled after dropping to almost nothing in the second quarter. It remains a smaller category than attachments. Microsoft's point is that, whichever format is used, the recipient's experience is the same: a message on a managed work device sends them to a phishing page on a personal phone.
These are Microsoft's own detection figures, from its own telemetry. They describe what its protections saw, not the entire internet, and the report is a vendor analysis rather than an independent count. We use them as a reasonable indicator of direction, not as a precise measure of how common the technique is overall.
The salary-review campaign, step by step
The report singles out a campaign on 19 August 2026. Microsoft Defender Research says it targeted more than 10,000 users across more than 4,800 organisations, almost all in the United States, and that law firms and legal organisations were the main focus. The messages were sent over roughly 13 hours, between the afternoon of 19 August and the early hours of 20 August in UTC.
The email itself had the subject line "Salary Review" and no visible body text. The only content was a Word document, which Microsoft says impersonated the recipient's own organisation, used company branding and addressed the person by name. It claimed that compensation had been revised after a performance review, and it told the reader to scan a QR code to see the details.
Placing the address in an image was the point. According to Microsoft, this moved the phishing link out of the email body and into a picture, which gets past gateway controls that read text links but do not decode QR code images.
Why scanning the code was not the whole trick
Scanning the code reportedly led to a page imitating a OneDrive sign-in, hosted on an unrelated domain. The page showed a short code and told the reader to copy it and continue to Microsoft's real sign-in portal. Microsoft describes this as device code phishing. The attacker's system had already asked Microsoft for a code, and when the victim entered it on the genuine Microsoft page and signed in, a valid session token went to the attacker.
According to Microsoft, that token gave persistent access to the account without exposing the password and without triggering multifactor authentication again. A countdown on the page pushed people to act before thinking. We are describing Microsoft's analysis here, and the report says scanning "would have" led to the page, so it is not a count of victims. The article does not state how many people scanned the code or how many accounts were compromised.
What this means if you scan QR codes
The advice that follows is general good practice and is consistent with what we say in our guide on whether QR codes are safe:
- Treat a QR code in an email or document as a link. If an unexpected message asks you to scan something to view pay, a shared file or a security alert, open the real service yourself instead.
- Check where the code leads before you act. Our guide to scanning a QR code on iPhone and Android explains how to read the address your phone shows.
- Be wary of any page that asks you to copy a code into a sign-in. A genuine request to sign in rarely begins with a stranger's page giving you a code to carry elsewhere.
- Report it. Tell your IT or security team, because the same message probably reached colleagues too.
Printed codes carry the same logic. We covered altered event posters earlier this month, where a real poster's design was kept and its code swapped. The delivery changes, but the habit of trusting a code because of where it appeared is the weakness.
What a dynamic QR platform cannot fix
Here is the plain version. A dynamic QR platform controls the codes created in its own accounts. It cannot see, scan or disable a code that an attacker generates with another tool and drops into a Word document. It also cannot inspect your mailbox, block an email or stop a device code sign-in. Nothing in our system, or anyone else's QR product, does those jobs. That is the work of email security, identity controls and phishing-resistant sign-in, which Microsoft's report discusses at length.
Anyone who tells you QR software alone solves phishing is selling something. The defence for recipients remains verifying the destination, and for organisations, it sits in the email and identity layers.
What a dynamic QR code gives a genuine sender
The useful angle for a business is different. If you send genuine QR codes, you want recipients to be able to tell them apart from forgeries, and you want to notice quickly if something unusual happens to your own codes. A dynamic QR code generator helps in specific ways, all of which we can confirm in our own product.
A predictable destination
On paid plans, custom domains for QR codes let your codes open pages on your own address rather than a generic short-link domain. This does not prevent forgeries. It gives you a rule you can teach staff and customers: our codes only open pages on this domain. If you ever send a QR code, keep that rule consistent, and never use a code to ask anyone to sign in or hand over a verification code.
Control after printing or sending
Because the code is dynamic, you can pause it or re-point it in seconds, and the change applies to every printed copy. If a mailing goes wrong, or you need scans to land on a notice about an impersonation attempt, nothing needs reprinting. Codes also have no expiry unless you set one, and free-plan codes stay live.
Evidence about your own codes
Each code has a scan log with country, city and device, and bot filtering separates automated previews from people, which matters because security tools often fetch links automatically. You can read more on bot filtering for scan analytics. Pro plans add scan alerts and webhooks, and paid plans can export logs as CSV. You can also set a country restriction, a password or an expiry date on a single code. None of that reveals an attacker's code. It does show unexpected behaviour on yours, such as scans from a country you never targeted. For the measurement side, our QR code tracking and scan analytics page and the pillar guide How to Track QR Code Scans explain what a scan log can and cannot tell you.
The wider lesson
Falling numbers are good news, but they describe averages, not the person who receives a convincing salary notice with their name on it. The August campaign shows how little an attacker needs: a believable pretext, a picture the email filter cannot read, and a phone that sits outside the company's defences. Volumes can rise again as quickly as they fell, and Microsoft itself notes that other techniques recovered while QR code attacks declined.
If you want to see how destination, scan logs and pause controls work before deciding what you need, you can create a QR code for free. For the numbers in this article, rely on Microsoft's report, and treat our commentary as one organisation's view of what it means for people who print and send codes.
Sources
All figures and campaign details come from Microsoft Threat Intelligence, Email threat landscape: Q3 2026 trends and insights, published 5 November 2026. Product statements reflect our own plan documentation at the time of writing.