Coruna Exploit Kit Hits Dutch Sites in Zero-Click Attack

Coruna

For three windows in early November, readers of more than a dozen Dutch news and sports sites loaded pages that tried to break into their phones. Nobody had to click anything. According to the publishers, the malicious code arrived inside an advertising tag, and the Coruna exploit kit was aimed at iPhones that had not been updated. Visitors on a current device saw an ordinary page. Visitors on an older one may have been somewhere else entirely.

The disclosure came from the publishers themselves rather than from a security vendor, which is unusual and useful. Dutch football site VI.nl, GeenStijl, Bright, Autoblog, Scientias and several magazine titles all used the same advertising partner, all carried near-identical notices, and all told readers roughly the same thing. An unauthorised third party had reached into an ad platform's management system and changed a tag. The result was a drive-by attempt to exploit unpatched iPhones through a Coruna chain.

This article sets out what the publishers said, what the Coruna exploit kit actually is, why the patch dates explain almost the whole story, where the public account contains an unresolved contradiction, what was and was not collected, and what a person with an older iPhone should do now. Where a detail comes from the publishers, it is attributed to them. Where it comes from Google's threat intelligence group, it is attributed there. The distinction matters, because no independent laboratory has published an analysis of the Dutch samples.

What the publishers disclosed

The notices describe a specific and limited compromise. The wording was near identical across the sites, and the version published by VI.nl is representative: an unauthorised third party gained access to the management system and made unauthorised changes to it, which allowed malicious code to be added to an advertising tag used on the site and others. The code has since been removed and the vulnerability in the advertising system has been fixed.

The publishers also described a data collection layer that did not depend on any exploit. Visitor data was obtained without anyone clicking anything, including IP addresses, pages visited, language settings and information about the operating system. That data went to a server controlled by the attackers. This part is straightforward tracking exfiltration, and it affected every visitor to an affected page during the window, regardless of device.

The timing was reported with unusual precision. The code was active from roughly 09:50 on 1 November until 10:07 on 3 November, with a second insertion later that day between 12:06 and 17:30. A further unauthorised change was recorded on 11 November between 15:34 and 15:50. That is three separated windows rather than one continuous exposure, which is consistent with an attacker testing access and returning to it.

The affected titles named in reporting include VI.nl, VoetbalPrimeur, VoetbalNieuws, Bright, Scientias, Autoblog, Men's Health, Women's Health, Runner's World, Bicycling, GeenStijl and Autoreview.nl. One tally puts the total at 14 Dutch sites plus one Belgian site. The common factor is the advertising partner rather than the subject matter, which ranged from football to science to fitness magazines.

What the publishers did not do is name a security firm or publish technical indicators. Their notices refer only to a preliminary technical investigation. No national cyber security centre, data protection authority or telecoms regulator had published a notice at the time of writing, and the advertising company did not respond to requests for comment according to Dutch reporting.

What Coruna actually is

Coruna is an iOS exploit kit, meaning a packaged set of exploits designed to take a phone from a hostile web page to running code on the device. Google's Threat Intelligence Group published a detailed analysis in March 2026 after recovering the full kit, and that analysis remains the best public description of what it can do.

The Coruna kit contained five complete iOS exploit chains and 23 individual exploits, covering iOS 13.0 from September 2019 through 17.2.1 from December 2023. Google described the surrounding framework as extremely well engineered, with the exploit pieces connected through shared utility modules. The name Coruna was not assigned by researchers. Google learned it from a debug build that an operator deployed by mistake, leaving the internal code names in the clear.

The technique is what makes Coruna dangerous. A page loads a small JavaScript framework that fingerprints the device, working out the model and the exact software version. It then loads the exploit chain written for that version, beginning with a remote code execution bug in the web rendering engine, followed by bypasses for pointer authentication, the sandbox, and kernel protections. Several of the Coruna exploits reuse techniques seen in earlier commercial spyware operations, including vulnerabilities that appeared in the Operation Triangulation campaign that Kaspersky documented in 2023.

The Coruna kit is also careful about who it attacks. Google found that Coruna bails out if the device is in Lockdown Mode or if the user is in private browsing. The binary payloads are encrypted with a per blob key and packaged in a custom format, and resource URLs are derived from a hashed cookie, so a casual scan of a server does not reveal the exploit set. This is not a hobbyist toolkit, and the Coruna code base is far more capable than a typical web skimmer.

The ending payload is where the money is. Rather than behaving like surveillance software, the implant that Google recovered steals financial information. It can decode QR codes from images on disk and it scans text for recovery phrase patterns and keywords such as backup phrase or bank account, sending what it finds back to a command server. It also loads additional modules that target specific cryptocurrency wallets, with a list that includes MetaMask, Trust, Phantom, Uniswap, TON wallets and several others.

Coruna has been used by more than one operator. Google tracked it from a customer of a surveillance vendor, to watering hole attacks against Ukrainian users by a group it calls UNC6353 and describes as suspected Russian espionage, to broad campaigns run by UNC6691, a financially motivated group operating from China. How the kit moved between them is unclear, which is itself the point: advanced exploitation capability now circulates.

The patch dates explain almost everything

The most useful fact for a worried reader is also the least dramatic. Every vulnerability in the Coruna chain has a fix date, and those dates are well in the past. The kit is not effective against current software, and the exposure in this campaign fell entirely on devices that had not been updated.

Apple shipped the relevant fixes on a clear schedule. Version 18.7.3 arrived on 12 December 2025, and versions 15.8.7 and 16.7.15 arrived on 11 March 2026. The targeted range described by the publishers lines up exactly with those boundaries: iPhone XS and XR and newer running 18.7.2 or earlier, and older iPhones running 15.8.6 or 16.7.14 or earlier. A device one version newer was not in the target set.

Apple's own advisory language is unusually direct. The notes accompanying version 16.7.15 state that the fix associated with the Coruna exploit was originally shipped in iOS 17.2 on 11 December 2023. In other words, the vulnerability that the Coruna chain used against that branch had been public and patched for more than two years before this campaign. The Coruna chain is not a zero-day against a current phone. It is a well engineered tool for reaching devices that stopped receiving updates.

That reframes the risk. The people exposed were not careless users who ignored a prompt last week. Many were running phones that had reached the end of their supported life, including the iPhone 6s, the iPhone 7 and the first generation SE, all of which the publishers named as targets. For those devices the security update may not exist at all. The honest advice for someone in that position is not to update but to stop using the device for anything sensitive.

For newer phones the advice is simpler. The iOS 18 chain that Google documented runs through a JavaScriptCore bug tracked as CVE-2025-43529, a graphics layer bug tracked as CVE-2025-14174, two kernel issues tracked as CVE-2025-43510 and CVE-2025-43520, and a loader bypass tracked as CVE-2026-20700. All but the last were fixed in 18.7.3. A device on a current release is outside the range this campaign targeted.

An unresolved contradiction about browsers

There is a genuine disagreement in the public record, and it should be reported as a disagreement rather than smoothed over.

The publishers told readers that the browser did not matter, listing Safari, Chrome and the built in browser as equivalent paths. That is what you would expect from a web delivered exploit, and it is what a reader needs to know if the advice is simply to update.

A teardown by Zimperium's research team describes something narrower about how the Coruna loader behaves. In its analysis of the Coruna loader, the kit verifies that the device identifies as mobile Safari and rejects alternative iOS browsers. If that behaviour held in this campaign, then a visitor using Chrome on an old iPhone would not have been exploited even though the page still loaded.

Both statements can be wrong in part and right in part. The publisher notice was written for a general audience and may have simplified. The Zimperium finding describes the kit's own logic as observed in a different deployment, and no laboratory has published hashes or domains for the Dutch samples to confirm that the same build was used. Until someone publishes that comparison, the safe reading is that a vulnerable device was a target, and that the browser question is open.

The practical consequence is small, because the advice does not change. A person on an unpatched phone should update or stop using the phone for sensitive work, and should not take comfort from having used a particular browser.

What was collected, and what is not known

Two different kinds of loss are in play here, and mixing them produces bad advice.

The first is certain in outline and uncertain in detail. Visitor data went to an attacker server: IP addresses, pages viewed, language settings and operating system details. That is a privacy harm rather than a device compromise, it applied to everyone who loaded an affected page in the windows, and its scale is not public.

The second is the one that matters most and is explicitly unresolved. The publishers said the code was used to try to exploit vulnerabilities on certain iPhones, and that a successful attack could have exposed stored passwords, app login credentials, messages, photos, contracts and cryptocurrency wallet keys. They also said it has not been established whether private data was stolen or whether any individual phone was successfully compromised. The investigation was described as preliminary and continuing.

That is an honest position and it should be repeated rather than upgraded. No victim has reported a compromise, no forensic report has been published, and the presence of a Coruna exploit attempt in a page is not evidence that the exploit succeeded. The difference matters because the response differs: a person whose phone was merely targeted needs to update and rotate credentials, while a person whose phone was compromised needs to treat the device as hostile.

Anyone who wants to check the boundaries of their own exposure can compare the three windows against their browsing history and their device version. A visit on a current release during those windows carries little risk. A visit on an affected version during those windows deserves the full response described below.

Who was selling the ads

The common thread across the affected titles is a single advertising partner. Dutch reporting names the company as NextDay Media and describes it as managing advertising for around 150 websites that together reach roughly 12 million visitors a month. That scale is why a single compromise of a management system reached a dozen publishers at once.

The mechanism described by the publishers is not a sophisticated server exploit. Someone obtained access to the management system of the ad platform, which is the administrative interface where campaigns and tags are configured. With that access, adding malicious code to a tag is a configuration change rather than an intrusion into each publisher. The publishers were downstream of the compromise, not its cause.

This is the same structural weakness that shows up in every third-party script incident. A publisher delegates part of its page to a partner, the partner's administrative layer becomes the target, and the compromise propagates to everyone who trusts the partner. Our coverage of the Rank Math PRO backdoored update describes the same pattern in a different part of the stack, and our earlier reporting on sustained fake tool campaigns shows how consistently attackers prefer the supplier rather than the target.

Advertising systems are a favourite because of what they are built to do. They accept code from many parties, they serve it to every visitor, and they change constantly. A tag that looks normal today can point somewhere else tomorrow, and publishers rarely have the tooling to diff what an ad partner delivers against what it delivered last week.

What to do with an older iPhone

The publishers' advice is specific and worth following in order.

Update the device to the current release for its model if an update exists. Where no update exists, the phone should not be used for banking, email, messaging or anything holding a wallet. Apple's Lockdown Mode is the supported mitigation for devices that cannot be updated, and it is effective here for a concrete reason: Google found that the Coruna kit bails out when it detects Lockdown Mode. That is a rare case where a defensive feature is documented to defeat a specific Coruna chain.

Change passwords that were stored on the device, and do it from a different trusted device rather than from the phone in question. Turn on two factor authentication wherever it is offered. Watch for unusual login attempts and for phishing messages that reference the incident, because a public scare is a good pretext for a fake notice. If a cryptocurrency wallet was installed on the phone, move the funds to a new wallet with a fresh recovery phrase generated on a device that was never exposed. Do not reuse the old phrase.

One control that people reach for does not appear to help much here. Zimperium's analysis notes that domain blocklists offer no protection against first party scripts, which is what an injected ad tag becomes from the browser's point of view. A content blocker may reduce advertising exposure generally, but it is not a verified mitigation for this campaign and should not be treated as one. The verified mitigations are the update and Lockdown Mode.

For readers thinking about mobile hygiene more broadly, our assessment of a mobile safe browsing feature looks at what these tools do and do not cover, and our piece on keeping credentials away from a browsing session covers the account hygiene side of the same problem.

Questions readers are asking

What is the Coruna exploit kit? It is a packaged set of iOS exploits that Google's threat intelligence group documented in March 2026. The Coruna name came from the kit's own developers rather than from researchers. It contained five full exploit chains and 23 exploits covering iOS 13.0 through 17.2.1, and it was named by its own developers.

Did the Dutch campaign definitely use Coruna? The publishers attributed it to Coruna, and relevant fixes in Apple's advisories are described as associated with the Coruna exploit. No independent laboratory has published hashes or infrastructure for the Dutch samples, so the attribution rests on the publishers and on Apple's advisory language rather than on a public technical report.

Did anyone have to click to be attacked? No. The malicious code was delivered inside an advertising tag and ran when the page loaded. Visitor data collection happened for everyone on an affected page during the windows, with no interaction at all.

Was any iPhone actually compromised? The publishers said it has not been established whether private data was stolen or whether any device was successfully compromised, and described the investigation as preliminary. An exploit attempt in a page is not proof that it succeeded.

Which iPhones were targeted? Devices on versions at or below 18.7.2 for the XS and XR generation and newer, and at or below 15.8.6 or 16.7.14 for older models. The publishers named the iPhone XS, XR, 6s, 7 and first generation SE. Devices on 18.7.3 or later were outside the target range.

When was the malicious code active? From about 09:50 on 1 November to 10:07 on 3 November, again on 3 November between 12:06 and 17:30, and on 11 November between 15:34 and 15:50.

Sources

[fluentform id="2"]