The Executive by MaxwellCTI Digest: November 11, 2026
The Executive by MaxwellCTI Digest: November 11, 2026
Get the full story at intel.maxwellcti.com.
Microsoft researchers demonstrated a method to backdoor AI agents, threat actors used AI tools for cloud breaches and infrastructure exploits, and France blamed Iran for sabotaging supermarket freezers.
Novel techniques
Microsoft's Lazarus Method Backdoors AI Agents via Few Neurons
Microsoft researchers demonstrated the Lazarus AI backdoor, which plants a hidden backdoor in an AI model by editing fewer than 10 rarely used neurons. This method edits neurons directly rather than using poisoned training examples to force an AI agent to execute commands with its own permissions. Potential attackers include insiders, individuals with stolen accounts, those with access to build pipelines, or publishers of modified models on public hubs.
Threat Actors Exploit Unpatched Cisco BroadWorks and Openwave Flaws
A Chinese-speaking threat actor used an autonomous AI intrusion harness to exploit a Cisco BroadWorks vulnerability in a campaign targeting a European telecommunications provider and at least 13 of its customers. This method was used to steal credentials, enumerate assets, and harvest device and subscriber billing records.
Stolen AI keys lead to $740,000 in fraudulent railway charges
Threat actors exploited vulnerabilities in Railway's build infrastructure and OIDC binding to steal credentials and customer authentication secrets, resulting in $740,000 in fraudulent AI usage. The attack involved AI-powered agents using abliterated models and a harness similar to ARTEX. This incident is part of a campaign targeting at least nine AI developer platforms.
Hackers Weaponize Artex AI Tool for Automated Cloud Breaches
Threat actors are using Artex, an open-source AI penetration-testing framework, to execute automated credential attacks and cloud breaches at high speeds. This tool allows for autonomous attacks, such as moving from scanning GitHub secrets to exfiltrating data from an Azure Key Vault in five minutes. Nation-state and financially motivated actors are using the framework to brute-force passwords, dump databases, and abuse APIs.
Russia-linked Turla Deploys Cereal Toolset Against Kazakhstan and Romania Exchange Servers
The Russia-linked threat group Turla deployed a modular .NET toolset called Cereal to intercept emails from Microsoft Exchange servers at government organizations in Kazakhstan and Romania. This toolset avoids external command-and-control infrastructure by co-opting the compromised servers to manage operations and exfiltrate data. It uses the .NET native image cache to run inside legitimate processes and executes most tasks in-memory.
Business
СберСпасибо (SberSpasibo) - 3,288,711 breached accounts
Data allegedly taken from SberSpasibo, the loyalty program of Sberbank, was published online in March 2023. The breach involved 48 million user records, including 3.3 million unique email addresses, phone numbers, dates of birth, and loyalty program details. Bank card numbers were also included, stored as unsalted SHA-1 hashes.
Japanese furniture giant Nitori says hackers stole 37.6M customer records
An attacker accessed the network of the Japanese furniture retailer Nitori and may have stolen 37.62 million customer records. The breach potentially exposed personal and corporate member details, as well as order information. Nitori warned that this compromised information could be used for phishing and scamming attempts.
YouHack Breach Exposes Over 7 Million Account Credentials
The Russian-language hacking forum YouHack experienced a data breach that exposed 7.2 million unique email addresses. This included 107,000 forum members whose usernames, IP addresses, and mostly unsalted MD5 password hashes were compromised.
France Blames Iran for Cyberattack on Supermarket Freezers
Shahid Kaveh hijacked refrigeration controllers at Danfoss to sabotage food storage, causing approximately 20 fires. France's Ministry for Europe and Foreign Affairs described this as the first offensive cyber campaign on French soil carried out purely for sabotage.
Critical vulnerabilities
- Apple iMessage and Brotli (CVE-2026-111497, CVE-2026-113960, CVE-2026-113961): attackers can join group chats, access previous messages, or exploit a compression library flaw, and CVE-2026-111497 is exploited; Apple has released updates. (story)
- Apple iOS (CVE-2026-113963): an app can access sensitive user data; Apple has released security updates.
- Barracuda Networks Email Security Gateway (CVE-2026-116606, CVE-2026-111560, CVE-2023-7102, CVE-2023-2868): attackers can obtain root access or execute arbitrary code, and they are exploited; Barracuda Networks has released patches.
- Microsoft Office Outlook (CVE-2026-124194): unauthenticated attackers can steal NTLMv2 account hashes, and it is already exploited; Microsoft has released a patch. (story)