Editorial illustration for Supermarket sabotage turns refrigeration controls into physical risk

1. Supermarket sabotage turns refrigeration controls into physical risk

Connected refrigeration attacks destroyed food, disrupted hundreds of supermarkets and were linked in reports to at least 15 fires. Systems serving around 700 French grocery stores were remotely pushed into defrost mode. Temperatures reached 50 Celsius for about six hours. Similar activity was reported in Belgium, Luxembourg, Sweden and other European countries. French authorities attributed the October sabotage to a cyber unit affiliated with Iran’s Islamic Revolutionary Guard Corps. U.S. agencies later warned healthcare, food, chemical and transportation organizations about similar cold-chain exposure.

What this means: Temperature controls protect food, medicine, vaccines, blood, tissue and chemicals. That makes these incidents a safety and continuity problem as well as a technology problem. For leaders: place refrigeration systems in business continuity, safety and geopolitical risk planning. For defenders: find exposed controllers, remove unnecessary remote access, strengthen authentication and test manual overrides. Watch for unexpected temperature changes in controller alerts and unfamiliar remote sessions in network or access logs. Exposure is highest where old firmware, shared credentials or direct internet access remain in place. The larger lesson is that access to ordinary connected equipment can create immediate physical and financial damage.

Recommendation: Inventory every connected temperature-control system and secure or remove its remote access immediately.

Read the related Daily story →

Editorial illustration for Management servers fall within hours of public exploit release

2. Management servers fall within hours of public exploit release

N-central servers were compromised within hours of public exploit code appearing, giving attackers root access to a trusted management layer. Researchers identified at least 21 affected production servers, including 19 running the latest security update. The observed attackers installed cryptocurrency miners. Exploitation requires the Legacy API to be enabled. Earlier in the week, hosted environments were taken offline while fixes were prepared, and on-premises customers were told not to reconnect until updated.

What this means: A compromised N-central server can sit close to many devices and customer networks. For leaders: assume management infrastructure has a wider potential impact than an ordinary server. For defenders: disable the Legacy API and investigate affected servers as potentially fully compromised. Watch for unexpected mining processes or services in host telemetry and unusual Legacy API activity in web or application logs. The observed payload was a miner, but root access could support more damaging actions. The larger lesson is that public exploit code can reduce the available response time from days to hours.

Recommendation: Disable the Legacy API and treat every matching compromise indicator as evidence of full server takeover.

Read the related Daily story →

Editorial illustration for Provider access helps attackers steal about €100 million from a bank

3. Provider access helps attackers steal about €100 million from a bank

Attackers used access routed through a service provider to execute fraudulent transactions from Crédit Mutuel Arkéa’s own account. The bank acknowledged losses in the order of €100 million, while an outside estimate put the total as high as €118 million. Customer assets were not affected. Some systems were suspended, transfers were delayed and operations began resuming after checks and stronger controls. The attackers remain unidentified, and the final loss has not been publicly confirmed.

What this means: The theft shows how provider access can become direct financial exposure for a customer. For leaders: assess vendors by the transactions they can enable, not only the data they can view. For defenders: monitor third-party sessions with the same urgency as internal privileged activity. Watch for unusual provider logins in access records and abnormal transfers in transaction-monitoring systems. Independent approval and rapid suspension controls can limit the damage from a compromised provider. The larger lesson is that outsourced access never outsources accountability.

Recommendation: Require independent approval, transaction limits and rapid suspension controls for every provider that can move funds.

Read the related Daily story →

4. Stolen GitHub access exposes code pipelines and cloud secrets

Stolen GitHub access put code pipelines and cloud environments at risk across a large campaign. An attacker probed more than 60,000 repositories, and roughly 4,800 compromised accounts submitted pull requests to test write access. Secrets were confirmed stolen from at least 1,835 targets. The malware searched for repository secrets, cloud credentials, GitHub tokens and AI service keys. It also reverted changes and deleted workflow records to hide activity.

What this means: Valid development tokens can bridge source code, automation and cloud infrastructure. Where permissions allowed, the attacker could seek cloud persistence after stealing credentials. For leaders: treat developer and automation identities as privileged enterprise access. For defenders: revoke affected tokens, rotate reachable secrets and investigate unauthorized cloud changes. Watch for suspicious pull requests and gaps in workflow histories inside GitHub, plus unfamiliar sessions in cloud audit logs. The larger lesson is that trusted development access can become a path from one repository to wider cloud control.

Recommendation: Revoke affected tokens, rotate every accessible secret and investigate unauthorized repository and cloud activity.

Read the related Daily story →

5. Outlook attacks expose authentication data after a malicious file is opened

Outlook users can expose authentication data after opening a specially crafted message file. Microsoft confirmed active exploitation, and the weakness was added to the catalog of known exploited vulnerabilities. Outlook can connect to an attacker-controlled network location without displaying its usual external-content warning. That connection may expose the user’s NTLMv2 authentication hash. The Preview Pane is not an attack path, so the victim must open the malicious file.

What this means: An exposed authentication hash may help an attacker impersonate the affected user. The moderate severity label does not reflect the urgency created by confirmed exploitation. For leaders: ensure Office updates are included in emergency patch decisions rather than relying only on Windows status. For defenders: deploy the relevant Outlook update and verify actual fixed build numbers. Watch for unexpected outbound authentication attempts in network or endpoint logs and reports of users opening unusual message files. The larger lesson is that exploitation evidence can matter more than a score viewed without context.

Recommendation: Patch affected Outlook installations and verify fixed build numbers rather than trusting deployment reports alone.

Read the related Daily story →

6. Autonomous AI scales intrusions across strategic organizations

A likely small team used an autonomous AI framework to run 25 operations against at least 19 organizations. Unauthorized access or data retrieval was confirmed at 10 targets. Semiconductor, defense-policy and civil-society organizations were among those targeted. The framework planned and executed tasks with open-source tools and commercial AI models, but it exploited familiar weaknesses rather than novel ones.

What this means: Autonomous orchestration can increase the number and duration of intrusion attempts without improving the underlying techniques. Individual operations lasted from 20 minutes to more than 50 hours. For leaders: expect exposed services and weak identities to be tested more persistently and at greater scale. For defenders: remove repeatable paths such as weak credentials, open registration and exposed cloud interfaces. Watch for repeated registration or login attempts in identity logs and sustained bursts against cloud interfaces in audit records. The larger lesson is that AI changes attacker economics before it changes fundamental defensive priorities.

Recommendation: Reduce exposed services and weak identity paths before autonomous tools can test them repeatedly.

Read the related Daily story →

7. Forgotten cloud trust lets attackers steal customer data and secrets

Customer data and secrets were stolen after attackers used compromised credentials linked to Qubole’s Amazon Web Services account. The campaign affected more than a dozen organizations from October 14 through October 26. Attackers assumed trusted customer roles, enumerated resources, read secrets and environment variables, and downloaded accessible storage data. Some roles could read nearly all the data the attackers pursued. Dormant roles remained in place even after the platform was no longer used.

What this means: Cross-account roles can preserve powerful access after a business relationship or integration ends. That makes vendor offboarding a cloud security task as well as a procurement task. For leaders: require cloud identities and trust policies to be removed when services are retired. For defenders: revoke unnecessary roles, rotate reachable secrets and review every resource those roles could access. Watch for unexpected external role assumptions in CloudTrail and unusual secret or storage reads in cloud data logs. The larger lesson is that unused trust remains active access until someone removes it.

Recommendation: Remove obsolete third-party cloud roles and investigate every resource and secret those roles could reach.

Read the related Daily story →

8. Railway breach triggers $740,000 in fraudulent AI charges

Stolen platform secrets led to about $740,000 in fraudulent AI usage for Railway customers. Railway said attackers exploited two infrastructure weaknesses, obtained internal source code and reached credentials used to decrypt customer environment variables. At most 1.85% of workspaces, or roughly 100,000, were potentially affected, but that is an upper limit. Linked-account tokens were reset for 1,852 users. No evidence was found of customer repository or database access.

What this means: The incident shows how stolen platform secrets can create immediate financial harm. Attackers used AI agents during a roughly 26-hour contest with Railway’s engineers. For leaders: developer platforms can concentrate customer trust, operational access and cost exposure. For defenders: rotate potentially exposed secrets, revoke linked tokens and review connected services. Watch for sudden AI-provider spending in billing dashboards and token use from unfamiliar locations in identity or access logs. The larger lesson is that trusted automation can also enable machine-speed abuse.

Recommendation: Rotate exposed secrets, revoke linked tokens and review AI-provider usage for unauthorized charges immediately.

Read the related Daily story →

9. Malware hidden in 1.3 million TV boxes powers criminal networks

More than 1.3 million Android-based devices across over 200 countries were enrolled in a botnet through software embedded in their firmware. Most were television set-top boxes, although telehealth equipment, karaoke machines and digital signs were also observed. The devices could provide residential proxy access, deliver more malware and support fraud. The embedded software could reinstall after removal because it was built into the firmware. A coordinated disruption severed the control domains and effectively disabled the malware.

What this means: Owners could unknowingly have their bandwidth and internet addresses used for criminal activity. Low-cost connected hardware can therefore introduce supply-chain risk before it reaches the network. For leaders: hardware sourcing and certification decisions can create security consequences that last for years. For defenders: inventory unmanaged Android devices and isolate equipment that cannot receive trusted clean firmware. Watch for unexplained outbound proxy connections and bandwidth spikes in firewall or network telemetry. The larger lesson is that a cheap device can carry hidden infrastructure for someone else’s criminal business.

Recommendation: Isolate or replace unmanaged Android devices that cannot receive a trusted clean firmware image.

Read the related Daily story →

10. SonicWall remote access lets stolen credentials bypass MFA

Attackers with valid credentials can bypass authentication controls on vulnerable SonicWall remote-access services. The weakness is being actively exploited on specified Gen7 and Gen8 devices. It affects systems configured to use Security Assertion Markup Language authentication. That path can defeat related multifactor authentication protections. Fixed releases and a workaround are available for organizations that cannot upgrade immediately.

What this means: Credentials stolen through phishing or information-stealing malware may remain useful despite multifactor authentication. Exposure is limited to affected devices using the relevant authentication configuration. For leaders: do not assume a control label protects every alternate route into a remote-access service. For defenders: install the fixed release and review alternate authentication settings and recent sessions. Watch for unusual alternate-method logins in remote-access logs and unexpected configuration changes in appliance audit records. The larger lesson is that identity controls are only as strong as the systems enforcing them.

Recommendation: Upgrade affected SonicWall devices and investigate suspicious access through alternate authentication methods.

Read the related Daily story →

11. Rogue ScreenConnect tools spread through provider trust paths

Rogue remote-access clients were installed on at least 288 endpoints across 48 organizations through cloud-hosted ScreenConnect environments. All identified systems were targeted within about 90 minutes. Researchers traced likely initial access to versions of an extension that could accept a blank shared key. The malware also attempted to spread from customer devices to technicians by stealing session cookies. No successful exploitation was found against patched clients.

What this means: Remote-support platforms create trusted paths among service providers, technicians and customer systems. Those same paths can accelerate malicious activity when extensions or credentials fail. For leaders: third-party add-ons can expand risk even when the core platform is not compromised. For defenders: disable affected extensions, update clients and investigate every unauthorized remote-access installation. Watch for unfamiliar clients and toolbox commands in platform logs, plus unexpected technician session use in access records. The larger lesson is that management trust can speed both legitimate support and malicious spread.

Recommendation: Disable vulnerable extensions and investigate every unauthorized remote-access client as possible persistence.

Read the related Daily story →

12. Global malware takedown exposes 6.4 million stolen credentials

An international operation disrupted criminal infrastructure supporting ransomware, credential theft and fraud. Authorities and private-sector partners seized 235 servers and 257 domains associated with CastleRAT, Gootloader and StealC. Investigators identified approximately 6.4 million stolen login credentials. Police from nine countries participated in the action. No arrests or charges accompanied the latest operation.

What this means: The disrupted services helped other criminals gain access, steal data and support ransomware campaigns. Gootloader alone was linked to at least 25,000 infected computers and approximately $28.8 million in losses since 2018. For leaders: use the disruption as an opportunity to assess exposure rather than assuming affected accounts are safe. For defenders: check available victim notifications, reset exposed credentials and enable multifactor authentication. Watch for unusual sign-ins in identity-provider logs and new alerts tied to previously exposed accounts. The larger lesson is that infrastructure disruption helps, but organizations must complete recovery at the identity level.

Recommendation: Check for exposed credentials and rotate affected passwords before criminal infrastructure is rebuilt.

Read the related Daily story →

13. Chrome users face another browser weakness already used in attacks

Chrome users need another urgent update after active exploitation of a memory-handling weakness. The issue affects the browser’s XML component. Fixed desktop releases are available for Windows, macOS and Linux, with related security fixes also reaching Android. The release contains nine security corrections. This is the eighth Chrome weakness exploited during 2026.

What this means: Browsers process untrusted content throughout the working day, making delayed updates a broad endpoint risk. Automatic delivery can lag when browsers stay open or users postpone relaunching. For leaders: measure browser-version compliance rather than assuming automatic updates have completed. For defenders: force the stable release and verify installation across managed operating systems. Watch for old browser builds and pending relaunches in device inventories or browser-management consoles. The larger lesson is that silent update assumptions can leave widely used endpoints exposed.

Recommendation: Deploy the fixed Chrome release and require users to relaunch so installation can complete.

Read the related Daily story →

14. Virtual machine hosts face takeover through Xen Orchestra attacks

Organizations running Xen Orchestra face active attacks that can lead to administrator access and full takeover of XCP-ng hosts. The authentication bypass works against reachable management interfaces without requiring valid credentials. Fixed server packages are available in Xen Orchestra 6.9.1 and 6.10.0. A successful attack can ultimately provide root control over an XCP-ng node. Crafted authentication requests may not leave useful identifiers in normal application logs.

What this means: A compromised virtualization manager can expose virtual machines, hosts and entire managed pools. The management plane therefore belongs in the same critical tier as identity and backup infrastructure. For leaders: treat virtualization administration as infrastructure-wide control rather than another application. For defenders: update immediately, restrict the interface and conduct host-level investigation. Watch for unusual requests or new sessions in network and reverse-proxy logs, plus unexpected host changes in system telemetry. The larger lesson is that one management interface can become control over a much larger environment.

Recommendation: Update Xen Orchestra, restrict its interface to management networks and investigate hosts before rotating environment secrets.

Read the related Daily story →

15. Poland shuts down public services platform after cyberattack

Citizens, businesses and public institutions lost access to online government services after Poland temporarily shut down its ePUAP platform. Security teams detected and stopped a cyberattack, then disabled the service as a precaution. The government said infrastructure supporting key digital services remained secure. There was no public confirmation that attackers entered government systems or stole data. No restoration timeline was provided.

What this means: The disruption affected electronic submissions, government communications and other administrative procedures. Poland’s Trusted Profile identity service could also be temporarily affected. For leaders: plan for the public and operational consequences of a preventive shutdown even when an attack is contained. For defenders: do not shorten validation, recovery or communication work because damage has not been confirmed. Watch for unusual authentication activity in identity logs and service degradation in operational dashboards. The larger lesson is that defensive shutdowns can protect systems while still creating material public impact.

Recommendation: Review shutdown criteria, alternative service channels and public communication plans for essential digital platforms.

Read the related Daily story →