>
av.Amit VijayanCYBERSECURITY & RESEARCHLet\u2019s connect
Back to the library

Cyber news

Daily Cyber Threat Brief — November 17, 2026: Optimizely supply-chain attack plants malware on 178 major brands' sites

Thumbnail for: Daily Cyber Threat Brief — November 17, 2026: Optimizely supply-chain attack plants malware on 178 major brands' sites

Case file — Lead incident: Attackers exploited a vulnerability in Optimizely's Experimentation account signup flow to inject malicious JavaScript into customer projects, turning the experimentation platform's own content delivery network into a malware pipeline. Security researcher Scott Helme found 188 infected Optimizely script files belonging to 178 organizations after analyzing 1,611 scripts. BleepingComputer independently confirmed the malicious code loaded, or attempted to load, on the sites of Zoom, CNN, Fox News, BBC, Salesforce, GitLab, KLM, Air France, HSBC Insurance, Bitly, Change.org, and UNICEF USA, among others. The injected code pulled a second-stage payload from 1688360.cc/track.js that swapped cryptocurrency wallet addresses, skimmed payment card details, and served iOS exploit kits. Wallets embedded in the payload had received about $7,000 by November 16. Optimizely says it blocked the vulnerable path, removed the injected code, and notified all affected customers; by 21:52 UTC on November 16 only one of the 188 files — belonging to UK broadband provider TalkTalk — still carried the malicious code. Status: contained, with the investigation continuing and updates posted to Optimizely's status page.

Incident timeline

DateDevelopment
October 7, 2026Attackers use compromised third-party credentials to reach Uber data held by AI cloud provider Fireworks AI.
October 8–11, 2026Card-skimmer code injected into Fandango's checkout page intercepts payment card details entered by customers.
October 9, 2026Fireworks AI notifies Uber of the breach; Uber severs its data connections with the provider.
October 13, 2026Uber's investigators confirm hackers used stolen provider credentials to access the data.
October 30, 2026Fandango determines customer information may have been impacted by the checkout skimmer.
November 1, 04:33 UTCAn attacker begins a 44-hour intrusion into Barracuda's cloud environment using a cloud access key the company gave an AI vendor and never retired; the last recorded activity is November 2, 23:31 UTC.
November 4, 2026A Barracuda engineer spots unusual activity on their own account; the AI vendor contacts Barracuda about its own compromise the same day.
November 10–11, 2026Barracuda publishes a Trust Center advisory, followed by a CISO post-mortem describing the intrusion as machine-speed and attributing the tooling to the ARTEX AI framework.
November 13, 2026Fandango files breach notices with the Vermont and California attorneys general; Uber files notices with several state attorneys general.
November 16, early morningThe first Optimizely script injections are observed; researcher Scott Helme's team begins identifying the infected files.
November 16, 2026Coles Group discloses it is investigating a cybersecurity incident; BleepingComputer reports the Fandango skimmer; SecurityWeek and TechCrunch report Uber is notifying about 149,346 drivers.
November 16, 21:12 ETBleepingComputer reports the Optimizely supply-chain attack; by 21:52 UTC only one of the 188 infected files still contains the malicious code.

1. Optimizely supply-chain attack plants malware on 178 major brands' sites

A supply-chain attack on Optimizely turned one of the web's most widely used experimentation platforms into a malware distributor on Monday, November 16. Attackers exploited a vulnerability in the account signup flow for Optimizely Experimentation to inject malicious JavaScript into customer projects. Because Optimizely's scripts load directly into its customers' webpages to run tests and personalization, the poisoned files were served from Optimizely's own domain, cdn.optimizely.com, carrying a valid Optimizely certificate — so browsers and most security allowlists treated them as legitimate.

Security researcher Scott Helme, who runs the monitoring service Report URI, uncovered the attack: his team analyzed 1,611 Optimizely scripts and found 188 infected files belonging to 178 organizations. BleepingComputer independently confirmed, from scans recorded by urlscan.io, that the malicious script loaded or attempted to load on Zoom, KLM, GitLab, BBC, Scheels, CNN, American Girl, ManageBuilding, Smartsheet, Stamps.com, and Salesforce — and on Fox News, Mutual of Omaha, American Eagle, Air France, Change.org, HSBC Insurance, Bitly, IG, and UNICEF USA as well. Salesforce confirmed it is investigating the malicious modifications to Optimizely scripts used on some of its websites. Optimizely's own website was also found loading the same code.

BleepingComputer analyzed copies of the modified files and confirmed the injected code was designed to load a second-stage script from 1688360.cc/track.js. That payload combined three money-making functions: a cryptocurrency clipboard hijacker that swaps wallet addresses in copied text, a payment card skimmer, and the Coruna and DarkSword exploit kits targeting iPhones and iPads. Helme's investigation found the cryptocurrency wallets embedded in the payload had received approximately $7,000 in transactions as of November 16 — evidence the clipboard hijacking had already paid off. The earliest observed injection was in the early morning of November 16, and the campaign ran for roughly 17 hours before Optimizely's cleanup took hold: a mass removal between about 20:40 and 21:10 UTC cut the infected scripts from 169 to one, and by 21:52 UTC only a single file — belonging to UK broadband provider TalkTalk — still contained the malicious code.

Optimizely confirmed the incident to BleepingComputer, saying an unauthorized party exploited the signup-flow vulnerability to inject the code, that it has blocked the vulnerable path and removed the malicious code, and that it has notified all affected customers. Updates are posted to its status page. It remains unclear how long the vulnerability had been present, or whether the attack involved a compromise of Optimizely's own infrastructure, stolen credentials, or another method of access.

Investigation notes

Why this is the lead: a single platform flaw put malicious code in front of the visitors of 178 organizations, including some of the most-visited news, video, and retail sites on the internet — and the code arrived signed by a trusted vendor domain. Independent researcher Amirul Cyber has dubbed the campaign "Optimjack." The supply-chain pattern is the story: the victims did not get hacked; their vendor did, and their sites carried the consequences.

What is confirmed: the 188 infected files across 178 organizations and the 1,611 scripts analyzed (Helme, November 16); the urlscan.io confirmations and the 1688360.cc/track.js payload (BleepingComputer); the signup-flow vulnerability and the cleanup (Optimizely's statement to BleepingComputer); the roughly $7,000 in wallet receipts and the 21:52 UTC cleanup figure (Helme).

What is not known: how the attackers gained the ability to modify so many customer projects, how long the vulnerability existed, and whether any visitors' data was actually stolen — the $7,000 in wallet receipts shows the clipper worked at least once, but no count of affected visitors has been published. Optimizely confirmed in February 2026 that attackers had accessed some of its internal systems through a voice-phishing attack; no researcher has linked that incident to this one.

What site visitors should do: anyone who visited an affected site on November 16 and entered payment details or copied a cryptocurrency wallet address should check their card statements and verify any crypto transaction destinations. Site operators using Optimizely should confirm their projects are clean and review what their third-party scripts are permitted to load.

2. Uber notifying about 149,346 drivers after Fireworks AI breach

Uber is notifying close to 150,000 drivers that their personal and driver's license information was stolen in an October breach at one of its technology providers. In notices filed this week with several US attorneys general, including California and Iowa, Uber said the provider told it on October 9 that hackers had broken in and stolen internal credentials, which were then used to access Uber's data. The Oregon attorney general's notice puts the total at 149,346 people.

The stolen data includes drivers' names, addresses, dates of birth, and driver's license numbers — including images of licenses, according to SecurityWeek. The provider was later identified as Fireworks AI, a cloud platform that gives developers access to open-weight AI models; an Uber spokesperson confirmed to TechCrunch that the notices relate to an incident at Fireworks AI and deferred further questions to the vendor. Fireworks attributes its breach to a financially motivated cybercriminal group that steals tokens for accessing large language models; no group has taken credit for the data theft. Uber says it severed its data connections with the provider, launched an investigation, and by October 13 its investigators had determined that hackers used compromised provider credentials on October 7 to reach the data. It is offering affected drivers free credit monitoring and fraud assistance.

This is Uber's second known data breach this year, and it is the second story in this brief — after the Cursor incident covered on November 15 — in which the Fireworks AI breach has surfaced as the source of a downstream victim's stolen data. Fireworks itself has acknowledged its breach also affected account and billing information, some customer datasets, and incomplete models.

Investigation notes

The pattern to note: the attackers never touched Uber's own infrastructure. They stole credentials at an AI vendor and walked through a door Uber had opened for it. The Fireworks breach is now a multi-victim event: the vendor's own data, then Uber's drivers, with the full list of affected Fireworks customers still undisclosed.

What is confirmed: the 149,346 figure (Oregon attorney general notice, via TechCrunch); the October 7 credential compromise, October 9 notification, and October 13 confirmation (SecurityWeek, November 16); the Fireworks AI identification (Uber spokesperson to TechCrunch); the data types, including license images.

What is not known: how the Fireworks credentials were stolen in the first place, whether the number of affected Uber drivers will grow, and which other Fireworks customers had data exposed. Uber did not say how many drivers it is notifying in total, and did not respond to a follow-up from TechCrunch on Monday.

What affected drivers should do: enroll in the free credit monitoring Uber is offering and watch for identity-theft attempts built on the stolen license numbers, dates of birth, and addresses. A driver's license number plus date of birth and address is enough to open accounts and pass identity checks.

3. Fandango checkout skimmer stole payment card details for four days

Ticket seller Fandango Media is notifying customers that hackers stole payment card details straight from its checkout pages. Card-skimmer code injected into the checkout page intercepted card numbers, CVV codes, expiration dates, billing ZIP codes, and email addresses entered between October 8 and 11, 2026 — four days of purchases on the biggest online movie-ticketing platform in the US, which sells tickets for 31,000 cinema screens including AMC, Regal, and Cinemark.

Fandango's notice to customers says its investigation found "evidence that an unauthorized party could access payment card information associated with certain transactions on our checkout page from October 8-11, 2026," and that it determined on October 30 that customer information may have been impacted. The company filed breach notices with the Vermont and California attorneys general on November 13, and BleepingComputer reported the incident on November 16. Fandango says it has fixed the checkout page, investigated with outside cybersecurity specialists, and is not aware of any fraud using the stolen data so far. It has not said how many customers are affected. The exposed data is the full set needed to spend a card online; Fandango says names, government-issued identity information, Social Security numbers, and other account information were not exposed. It is offering two years of free IDX credit and dark-web monitoring, plus identity theft recovery services, with enrollment open until February 13, 2027.

Investigation notes

Why the timing matters: the skimmer ran October 8–11, the impact was confirmed October 30, and customers are hearing about it on November 16 — more than a month after the last exposed purchase. The window for watching statements has already been open for weeks, and the card details were live in attackers' hands throughout.

What is confirmed: the October 8–11 exposure window and the five data types (Fandango's customer notice, filed with the California attorney general); the October 30 determination date; the November 13 state filings; the fix and the monitoring offer, including the February 13, 2027 enrollment deadline.

What is not known: how many customers are affected — the Vermont notice counts 18 residents of that state, but no nationwide total has been published — how the skimmer got onto the checkout page, or whether the stolen details have been used. Fandango says it is not aware of fraud so far, which is not the same as confirming none occurred.

What Fandango customers should do: anyone who bought tickets on Fandango between October 8 and 11 should check card statements for unauthorized charges, consider replacing the card used, and enroll in the free monitoring before the February 13 deadline.

4. Coles investigates cybersecurity incident; no data theft found so far

Coles Group Limited, Australia's second-largest retailer, is investigating what it described as a "cyber security incident" that has not impacted the company's stores. In a statement issued on Monday, November 16, the company said it is working with external experts to secure its systems and has yet to find evidence that attackers stole customer, employee, or supplier data from compromised systems.

"Coles Group Limited is currently investigating a cyber security incident. Based on the findings of our investigation to date, we have not identified any evidence that customer, team or supplier data has been compromised or removed from Coles' systems," the company said. Coles is working with the Australian Cyber Security Centre and says it will continue to engage with appropriate agencies and regulators. Customers can shop as normal: stores, online outlets, Flybuys accounts, Coles Plus, Coles Plus Delivery, and QuiteLike subscriptions have not been affected. Coles noted that a small number of customers had click-and-collect issues at Liquorland outlets, but did not link those problems to the incident. The company has not said when it discovered the incident or whether a ransom demand was made, and no major threat group has claimed the attack. It apologized to customers and recommended regularly updating passwords with strong, unique ones and not reusing them across services.

Investigation notes

What is confirmed: the incident and the no-evidence-of-data-theft finding so far (Coles Group statement, November 16); the Australian Cyber Security Centre's involvement; stores and the named services unaffected.

What is not known: the nature of the intrusion, when it was discovered, whether data was taken, and who is behind it. "No evidence found so far" is an interim finding in an active investigation, not a clean bill of health — and the disclosure gives no technical detail at all, which leaves the scope open.

What Coles customers should do: follow the company's own advice — update the Flybuys and Coles account passwords to strong, unique ones — and watch for phishing that trades on the incident. No action is needed for in-store or online shopping, which Coles says is unaffected.

5. Barracuda: an unretired vendor key and an AI attack tool, 10 cloud accounts in 44 hours

Barracuda has published its post-mortem of a fast-moving intrusion into its cloud environment at the start of the month, and the cause is uncomfortably ordinary: a cloud access key it gave to an AI vendor for a past project and never retired. The vendor's development environment was compromised in October, and the key was sitting in a configuration file in the vendor's GitHub repository. The attacker used it to make the first call into Barracuda's cloud at 04:33 UTC on November 1, and within minutes was mapping what the key could reach and reading stored secrets — AWS access keys and Azure service principal secrets among them.

Each secret opened the next system: about ten cloud accounts, a data analytics system, a source-code access token, then Kubernetes clusters. The attacker decrypted infrastructure-as-code state files, pulled credentials from server metadata services and Kubernetes service accounts, and in under four hours pushed short-lived branches to a handful of repositories to trigger a build workflow that printed the build system's secrets into the job log. The last recorded activity was November 2, 23:31 UTC — about 44 hours after the first call. Detection came on November 4, when an engineer noticed odd activity on their own account and reported it; the vendor contacted Barracuda about its own incident the same day. Barracuda removed the container that evening, rotated credentials across every system involved, and later retired two internal automation servers with exposed credentials, with no sign the attacker used them.

The system that stores Barracuda XDR integration credentials — used to pull data from customers' Microsoft 365, Google Workspace, and endpoint tools — was accessed. Barracuda found no direct evidence those credentials were viewed, but says its logging was not detailed enough to rule it out, and it has contacted the customers it believes could be affected. The company matched the attacker's probe names, hunting routines, and machine-style resource labels — the letter "i" followed by digits, with no human-readable names — to ARTEX, the open-source AI penetration-testing framework the FBI warned about on November 13. Barracuda is blunt about the lesson in its own words: "That is on us." Customers who were not contacted directly need take no action; those who were should revoke and recreate integration credentials and review connected-system activity logs from November 1.

Investigation notes

Official anchor: Barracuda's Trust Center advisory of November 10, updated November 13, and the CISO post-mortem of November 11, reported by BleepingComputer on November 16. The ARTEX attribution rests on tooling fingerprints — probe names, routines, and resource labels — which establish how the attacker operated, not who it was.

What is confirmed: the unretired vendor key in the vendor's GitHub configuration file; the November 1–2 intrusion window and the roughly 10 cloud accounts reached; the November 4 detection; the XDR credential store access, with viewing neither confirmed nor ruled out.

What is not known: whether the vendor's GitHub repository was public or private, which cloud the key belonged to, and whether any customer XDR credentials were actually viewed or used. Barracuda believes the intrusion was the work of a financially motivated group.

What defenders should take from it: vendor access keys have a retirement date, not just an issue date — and a key left in a partner's repository outlives the project it was issued for. Audit every key ever handed to a vendor, contractor, or AI tool, and treat a partner's compromise as your own exposure until proven otherwise.

Sources

Keep exploring.

All articles