VulnCheck: 14 Copeland XWEB EVO flaws, 2 exploited, first 1 h 42 min after the NVD record (derived), 5 with no fix
VulnCheck says two of 14 flaws in Copeland's XWEB EVO refrigeration supervisor are exploited, the first 1 h 42 min after the NVD record went live (derived). Five reach the current firmware, and Copeland's bulletin says security patches ended on 16 March 2026. Read on 16 November 2026.
By Parminder Kumar Sharma · LinkedIn · · 18 min read
1 h 42 min from a one paragraph NVD record to the first recorded hit (derived)
VulnCheck's post of 13 November 2026 says the NVD record for CVE-2026-124014 went live at 20:18 UTC on 12 November as a one paragraph description, with no exploit and no public proof of concept. Its canary sensors then recorded the first hit at 22:00:05 UTC. The gap is 1 h 42 min 5 s (derived: 22:00:05 minus 20:18). The post itself says "an hour and forty minutes" and its timeline says 1h40. The NVD time is printed to the minute, so the interval is between about 1 h 41 min and 1 h 42 min, and this brief uses 1 h 42 min.
What that gap does not establish. It does not show that any controller was compromised: the post counts hits and probes on VulnCheck's own canary sensors, and it says the other 12 flaws have no hits in the wild. It does not say who sent the traffic, how that source learned of the flaw, or that the NVD record was the source of its knowledge. VulnCheck says the sending host reads as an opportunistic scanner, and that neither scanning source is evidence of the October French refrigeration campaign. It also says nothing yet about how many of the roughly 2,000 exposed XWEB EVO controllers are the legacy firmware that the two exploited flaws need.
The product is XWEB EVO, a Dixell refrigeration supervisor now sold under the Copeland (Emerson) name. VulnCheck disclosed 14 vulnerabilities in it, CVE-2026-124002 to CVE-2026-124015, and says each was taken to root on lab appliances rebuilt from the vendor's firmware. Nine affect only firmware from before a 2021 rewrite. Five reach the current 5.8.8 firmware, which Copeland's own bulletin says gets no more security patches since 16 March 2026.
Whose numbers these are. VulnCheck sells vulnerability and exploit intelligence, runs its own known exploited list (VulnCheck KEV) and was the CNA that assigned these 14 ids, so the first listing claim and every sensor count in its post are its own data and bear on its product. SecurityWeek's report of 16 November is a relay of the post. Copeland's bulletin is the vendor's own page. This site read all three on 16 November 2026, the SecurityWeek page being dated 06:05 ET (11:05 UTC). Anything about listings, patches or counts can change within the hour.
Fourteen flaws in two groups, and which two are being hit
VulnCheck groups the 14 by firmware generation. Nine sit on the pre-2021 firmware that VulnCheck calls the .lua era: its post says that class was removed in the 2021 rewrite, so it is fixed on the current branch, but lives on across the legacy fleet that it says is still most of what is exposed. VulnCheck prints no fix version for any of the nine and no count of exposed legacy hosts. Five survive the rewrite and reach 5.8.8. The lab firmware VulnCheck names is a native i386 version of 5.2.3.STANDARD for the legacy flaws and the ARM 5.8.8.STANDARD firmware under emulation for the credential path.
The two exploited flaws are both in the nine. CVE-2026-124014 is an unauthenticated write of the system configuration file through config.lua. CVE-2026-124010 is unauthenticated root command execution through a setuid-root CGI, which the post calls the most severe of the 14. SecurityWeek relays both points and adds that both targeted flaws affect only the older firmware. So the exposure that matters today depends on which firmware a device runs, a fact the post and the relay do not count.
Why this box. VulnCheck says it looked at XWEB EVO after the attack of 11 and 12 October 2026, in which, in its words, roughly 700 French supermarkets had refrigeration controllers forced into defrost overnight and in at least 15 stores the overloaded units caught fire. It says CERT-FR named four controller families as compromised, among them Copeland and Dixell XWEB, and that France attributed the campaign to an Iranian IRGC cyber unit on 9 November. SecurityWeek relays the attribution as France blaming Iran. Neither saved page is a French government text, and this site attributes the claim without adopting it. The post separately says the two scanning sources it saw are not evidence of that campaign. This site's earlier briefings cover the October wave: the first week and the second wave of 2 November.
The clock: 241 days of unsupported software, then a 1 h 42 min gap
Copeland's bulletin, which prints a month and year (March 2026), lists the end of sales as OCT-2023 and the end of software update as MAR-16, 2026. Among the activities that stop, it lists the release of security patches for identified vulnerabilities. From 16 March to the NVD record on 12 November is 241 days (derived). The bulletin and VulnCheck's account, taken together, leave no room for a vendor patch cycle on the five flaws that reach 5.8.8.
Intervals this site computed from times printed in the VulnCheck post, the Copeland bulletin and SecurityWeek. Every result is derived
| Interval | From and to | Derived result |
|---|---|---|
| End of software support to the NVD record | 16 Mar 2026 to 12 Nov 2026 | 241 days |
| End of software support to the post, then to SecurityWeek | 16 Mar to 13 Nov, and to 16 Nov | 242 days, then 245 days |
| Submission to Copeland PSIRT to the NVD record | 17 Oct to 12 Nov | 26 days |
| Submission to the PSIRT reply, then reply to the NVD record | 17 Oct to 10 Nov, then 10 Nov to 12 Nov | 24 days, then 2 days |
| NVD record to the first hit on CVE-2026-124014 | 20:18 to 22:00:05 UTC, 12 Nov | 1 h 42 min 5 s |
| NVD record to the first probe on CVE-2026-124010 | 12 Nov 20:18 to 13 Nov 03:41:20 UTC | 7 h 23 min 20 s |
| First hit on 124014 to the last of its 125 hits | 22:00:05 to 06:16 UTC | 8 h 15 min 55 s |
| Hits after the first hour from the first source | 125 minus 87 | 38 hits |
| The four probes on CVE-2026-124010 | 03:41:20 to 04:09:55 UTC | 28 min 35 s |
The disclosure window is printed oddly. VulnCheck's timeline says the 14 findings were submitted to Copeland PSIRT on 17 October 2026 with a 120 day disclosure window closing 2026-02-16, a date that falls before the submission. This site does not know which date was meant. Counting 120 days from 17 October 2026 gives 14 February 2027 (derived), and the 12 November publication came 26 days after submission, 94 days short of that (derived: 120 minus 26). The timeline then says Copeland PSIRT replied on 10 November, citing end of support, and left the call on whether the flaws duplicate XWEB Pro records to VulnCheck. The post prints no reply wording, and says nothing on whether Copeland agreed the 12 November date.
VulnCheck's own lab clock. The post prints the log of a lab run of an exploit module that its Initial Access Intelligence team wrote for CVE-2026-124002, one of the seven with no hits in the wild, stamped 11:43:46 at plus one hour, which is 10:43:46 UTC (derived). That is 9 h 34 min before the NVD record (derived). Beside it the post says VulnCheck publishes a new exploit every Monday in its Initial Access Community repository. The saved page does not say that this module has been published or when. It is a point about who held working details early, not about who sent the traffic.
Which flaws have a fix: nine on the legacy firmware, five with none coming
What each page says about fixes for the two groups. VulnCheck post of 13 Nov, SecurityWeek of 16 Nov, Copeland bulletin of March 2026, all read 16 Nov 2026
| Group | What the pages say | Not stated |
|---|---|---|
| Nine on pre-2021 firmware (124002, 124003, 124004, 124007, 124008, 124009, 124010, 124014, 124015) | VulnCheck: the class was removed in the 2021 rewrite, so fixed on the current branch. SecurityWeek: these only affect firmware that predates the rewrite. | A fix version. Whether any given legacy unit can be moved to the current branch. How many exposed hosts are legacy. |
| Five that reach 5.8.8 (124005, 124006, 124011, 124012, 124013) | VulnCheck: won't fix by the vendor's own policy. SecurityWeek: these five will not be fixed. Bulletin: vulnerabilities found after 16 Mar 2026 will not be fixed. | Any Copeland statement on these five ids. VulnCheck's wording is its reading of the bulletin policy and the PSIRT reply. |
| Severity scores | VulnCheck calls 124010 the most severe, in words. | Any CVSS score for any of the 14, in any page read. |
The two statements are consistent because they describe different groups: the nine are fixed in a later firmware generation, while the five are in that later generation and will not be fixed. The catch for an owner is that the first statement is not a patch for an old unit. VulnCheck's phrase is that the class was removed in the 2021 rewrite. This site's inference, labelled as such: a controller still on the legacy firmware is exposed to nine flaws, two of them under attack, unless its owner has moved it to the current branch, and Copeland's bulletin says updates for the whole family stopped on 16 March 2026. The pages read do not say whether a move to the current branch remains available.
Friendly labels doing the work of a control. Two phrases can be read as comfort. Fixed on the current branch does not fix a device that is not running it. Transition to XWEB PRO models, the bulletin's required action, points owners to a platform that VulnCheck says shares the Lua middleware with XWEB EVO. VulnCheck cites separate 2026 research by Claroty Team82 and a CISA advisory, ICSA-26-057-10, covering the Pro line only, and says a few of the 14 are plausibly the same bug on the EVO codebase, which it could not confirm because the Pro firmware is encrypted. Onyphe's survey, as the post cites it, counts 108 public Pro hosts. A supported label is not a clean one. This site's briefing on XWEB PRO controllers in a suspected spam botnet shows the Pro line has had its own exposure.
What each number counts, and what none of them counts
The post prints a first source with 125 hits through 06:16 UTC on 13 November, 87 of them in the first hour, and a second source with four probes between 03:41:20 and 04:09:55 UTC. VulnCheck says the first host is on a commodity hosting network, also sprayed a few unrelated recent CVEs, and has no botnet, named threat actor or ransomware group tied to it in VulnCheck's own indicator feeds. It calls the second an established opportunistic scanner, with 189 requests across nine CVEs in the previous 90 days, eight of the nine unrelated to XWEB. The four XWEB probes ran only an identity command, with no second stage payload or command and control indicator in the requests, per the post.
On exposure the post gives two counts that do not match: about 1,663 hosts for a Shodan query on a combined XWEB EVO fingerprint, and about 1,960 public XWEB EVO hosts against 108 for Pro in an Onyphe survey. The two differ by 297 (derived: 1,960 minus 1,663) because they are different surveys, and the post does not reconcile them. Onyphe's two numbers give a ratio of 18.1 EVO hosts per Pro host (derived: 1,960 divided by 108). SecurityWeek relays the Onyphe figure as nearly 2,000, and a ratio from that rounded number would be 18.5 (derived: 2,000 divided by 108), so this brief uses the post's own 1,960. VulnCheck also says its Target Intelligence has no entry for the end of support EVO firmware, so to that dataset it does not exist.
What the records establish and what they leave out. VulnCheck post, SecurityWeek and Copeland bulletin read 16 Nov 2026, plus this site's own check of the CISA catalogue
| Item | What a page states | Not stated |
|---|---|---|
| First hit on 124014 | 22:00:05 UTC on 12 Nov, one source, 125 hits to 06:16 UTC | Who runs it, how it learned of the flaw, whether any device was compromised |
| Second source on 124010 | Four probes, 03:41:20 to 04:09:55 UTC on 13 Nov | Whether it is the same actor as the first, any second stage |
| Canary sensors | VulnCheck calls them canary sensors | How they work, how they are exposed, what a hit is made of |
| VulnCheck KEV | Both added the day exploitation was first seen, ahead of any other KEV source, per VulnCheck | Any other list's dates. This site's check of CISA's catalogue (version 2026.11.13, 1,781 entries) on 16 Nov found neither CVE; it can change at any time |
| Exposed hosts | About 1,960 EVO and 108 Pro (Onyphe), about 1,663 (Shodan query) | How many of them run the legacy firmware, which the post says is most of what is exposed |
| Copeland | 17 Oct submission and 10 Nov PSIRT reply, per VulnCheck's timeline; March 2026 bulletin | Reply wording, any statement on the 14 ids, whether it agreed the 12 Nov date |
| The October campaign | VulnCheck: the scanning is not evidence of it | Whether the October campaign used any of the 14 flaws |
Where the records differ, left as differences
87 hits or 125 requests. The post prints both for the same source: 125 hits through 06:16 UTC, 87 of them in the first hour. SecurityWeek prints only the larger figure, as a single IP address sending 125 requests overnight. They count the same thing over different spans, so this is a difference of span and wording, not of data.
An hour and forty minutes or 1 h 42 min. The post's prose and timeline round the gap to forty minutes past the hour. The two printed times give 1 h 42 min 5 s (derived), or 1 h 41 min 6 s if the NVD record went live at the end of its minute (derived). Neither page says which clock or rounding it used.
Two Copeland bulletins. This site read the March 2026 bulletin only for this brief. It recommends an inventory of XWEB EVO installations, a transition plan to XWEB PRO and interim measures such as network segmentation, restricted access and enhanced monitoring. This site's earlier briefing of 2 November summarised a Copeland bulletin of 14 October 2026 as telling owners to remove XWEB EVO from the public internet; this site did not re-read that bulletin here. VulnCheck's advice, that an XWEB EVO must never touch the public internet, is the stronger of the statements read in this batch.
Who is blamed for October. VulnCheck states, and SecurityWeek relays, that France attributed the campaign to an Iranian IRGC cyber unit. Neither saved page is a French government text, and the CERT-FR alert and the foreign ministry declaration that the post cites were not read. Both pages agree on the grouping of the 14, the two exploited flaws and the Copeland end of support date, and SecurityWeek adds no data of its own beyond the post.
Listing dates. The post says CVE-2026-124014 was added to VulnCheck KEV the same day, 12 November, and CVE-2026-124010 on 13 November. A single phrase in its timeline reads as the same day for both. Each date is the day VulnCheck first saw the exploitation, per the post.
What the VulnCheck post prints
The post prints far more than this site repeats: request lines, code fragments from the recovered firmware, the way the lab images were unpacked, a fixed password, the daily code method and a log of its exploit module. This site does not copy any of it. What follows is the level of the post's own summary lines: the 14 ids with the component and weakness classes as printed, and the indicators the post lists for defenders, defanged. This site did not visit or look up any address.
The nine flaws on pre-2021 firmware, as the VulnCheck post of 13 Nov 2026 prints them. Effects are this site's plain words for the post's summary lines
| CVE id | Component and CWE classes as printed | Effect in plain words |
|---|---|---|
| CVE-2026-124010 | xwconsole.cgi; CWE-78, CWE-306, CWE-250 | One unauthenticated request to a setuid-root CGI runs a command as root. Called the most severe. Exploited per VulnCheck. |
| CVE-2026-124002 | login.lua, username field; CWE-78, CWE-306 | OS command injection in the login before the password check. |
| CVE-2026-124007 | login.lua, password field for hashed-mode accounts; CWE-78 | The same class of injection, through the password field. |
| CVE-2026-124004 | Administrator account; CWE-798 | A hardcoded backdoor password. This site does not print it. |
| CVE-2026-124003 | wlua Lua interpreter; CWE-250 | The interpreter is setuid root, so each injection above becomes a root injection. |
| CVE-2026-124008 | devices.lua template handler; CWE-78 | Authenticated command injection. |
| CVE-2026-124009 | backup.lua mount handler; CWE-78 | Authenticated command injection. |
| CVE-2026-124014 | config.lua; CWE-306, CWE-862 | Unauthenticated write of the system configuration file. Exploited per VulnCheck. |
| CVE-2026-124015 | users.lua; CWE-306, CWE-640 | A password reset that bypasses the session. |
The five flaws that reach firmware 5.8.8, as the VulnCheck post of 13 Nov 2026 prints them. Effects are this site's plain words for the post's summary lines
| CVE id | Component and CWE classes as printed | Effect in plain words |
|---|---|---|
| CVE-2026-124005 | Administrator code of the day; CWE-330, CWE-287 | The admin password is a secret free function of the device MAC and the date. |
| CVE-2026-124011 | sysinfo.lua; CWE-200 | Leaks the MAC and the clock to any unauthenticated caller. |
| CVE-2026-124012 | Administrator login; CWE-798, CWE-287 | Logs in with the daily password alone, no stored secret. |
| CVE-2026-124006 | Linux root password; CWE-798, CWE-1392 | Rebuilt daily as a fixed prefix plus that same daily code. |
| CVE-2026-124013 | api.lua import_preconfiguration action; CWE-78 | Authenticated command injection. |
The next block lists the product, the firmware versions VulnCheck says it tested and the component names it prints. A defender can use it to recognise what the advisory is about and to search request logs for the handler names. It is not a way to tell which firmware a device runs.
Product: XWEB EVO, the Dixell refrigeration supervisor sold under Copeland (Emerson)
Lab firmware, legacy flaws: 5.2.3.STANDARD (native i386)
Lab firmware, credential path: 5.8.8.STANDARD (ARM, emulated)
Components named: xwconsole.cgi, login.lua, config.lua, devices.lua, backup.lua,
users.lua, sysinfo.lua, api.lua (import_preconfiguration action), wlua
The post's list for defenders is two source addresses, two request patterns and two sensor rule names. The addresses are scanner hosts on commodity hosting networks according to the post, not victims, so they are useful for a log search and not as a block list on their own. A search that finds nothing proves little, because the post does not say these two hosts are the only ones, and the post's own words on the request patterns are an unauthenticated POST to config.lua and an unauthenticated GET to the xwconsole.cgi handler carrying a command parameter. This site does not print the parameter.
Source of the CVE-2026-124014 hits: 202[.]61[.]138[.]117 (AS152194, CTG Server Limited)
Source of the CVE-2026-124010 probes: 177[.]4[.]12[.]11 (AS140227, Hong Kong Communications International Co., Limited)
Sensor rule: VULNCHECK CVE-2026-124014 Copeland XWEB EVO Unauthenticated config.lua Configuration Write Attempt
Sensor rule: VULNCHECK CVE-2026-124010 Copeland XWEB EVO xwconsole.cgi Unauthenticated Root Command Execution Attempt
What to do, in order
The sources' own advice first. VulnCheck: an XWEB EVO must never touch the public internet, should sit behind a VPN, and its owner should plan the move to a supported platform. Copeland's bulletin: identify all active XWEB EVO installations, plan the transition to XWEB PRO, and meanwhile use network segmentation, restricted access and enhanced monitoring, in coordination with the security officer.
Take this with you
This site's suggestions, in the order worth doing
- Find out whether you own one. Ask the facilities team and the refrigeration contractor for every XWEB or Dixell supervisor in every store, cold room and depot. Do not scan for them.
- Ask the contractor which firmware generation each unit runs. VulnCheck says the pre-2021 generation is most of what is exposed and is where both exploited flaws live; the 5.8.8 generation has five flaws and no coming fix.
- Take any unit with a public address off the public internet this week, and put remote access behind a VPN, as both VulnCheck and the bulletin say.
- Search web server and reverse proxy logs back to 12 November for requests to the xwconsole.cgi handler and to config.lua from outside, and for the two source addresses in the technical section. Absence of hits is not proof of safety.
- If a unit was reachable from the internet, treat it as possibly compromised and follow the vendor and CERT-FR guidance, summarised in this site's earlier briefing of 2 November.
- Retailers and managed service providers: an end of support date is a replacement date, not an advisory date. Put the 16 March 2026 date into the asset register as the day the plan was due, and budget the move.
- When planning the move to XWEB PRO, ask the supplier for the Pro security advisory history and the current firmware status, because a supported label is not a clean one.
The question the records leave open
Two flaws are being hit, and both exist only on the older firmware. Onyphe counts about 1,960 public XWEB EVO hosts, VulnCheck says the legacy generation is most of them, and no page prints a count. Who knows which of those roughly 2,000 controllers run the legacy firmware, who will tell their owners, and what do they do now that the vendor's own bulletin ended security patches for the whole family on 16 March?
Sources
- PrimaryCode of the Day: Unauthenticated Root in Copeland XWEB EVO, dated 13 November 2026 on its page: the 14 CVEs, the canary sensor counts, the timeline. Read in full; VulnCheck sells vulnerability intelligenceVulnCheckaccessed 2026-11-16
- PrimaryProduct bulletin of March 2026: end of software support for the XWEB EVO family effective 16 March 2026, the activities that stop and the recommended actions. Read in fullCopeland Controlsaccessed 2026-11-16
- Reported byReport of 16 November 2026, 6:05 AM ET: a relay of the VulnCheck post, the statement that five issues will not be fixed, the nearly 2,000 exposed hosts. SecondarySecurityWeekaccessed 2026-11-16


