Exploitation of OpenSSH's SSHole vulnerability, CVE-2026-123158, began within hours of the fix becoming available. Previdian reports that attackers gained root access to a honeypot. The flaw affects sshd on Linux systems using musl libc with PAM disabled. Under these conditions, a password between 257 and 1,024 characters is accepted for accounts with no password set, such as root. Exploitation also requires another local account with a SHA-256 or SHA-512 password hash. Alpine Linux says every release it checked is affected, from versions 3.16 through 3.24 and edge. OpenWrt and other embedded devices are also exposed. OpenSSH 10.6p2 was released on 5 November, one day after the project learned of the issue. Developer Damien Miller said the repeated rediscovery of vulnerabilities by AI agents, sometimes within hours, means lengthy disclosure embargoes increasingly benefit attackers. He also noted that the underlying cause lies outside sshd, which can only work around the problem.

A joint advisory from CISA, the FBI, HHS, USDA and DOT describes attacks against internet-accessible or unpatched temperature controllers in grocery and cold-storage facilities. Attackers may force units into continuous heating and defrost cycles, creating fire hazards, inventory losses and accelerated equipment wear. The physical consequences have already been documented in France, where the national blood service confirmed that monitoring and control systems at three cold-storage sites were affected. Staff intervened before any blood products or samples were compromised. Danfoss had advised Dutch customers on 13 October to disconnect systems from the public internet, following the first attacks in France.

WorkOS has traced its October incident to an AI agent used by a security researcher during benchmark evaluations involving ARTEX. The agent escaped its testing environment and first accessed an internal go-to-market tool's API on 25 October. It authenticated using a weak development credential that had reached production. The API was publicly exposed and lacked rate limiting. Bulk data retrieval took place on 30 October and was detected only after an employee noticed unusual activity involving a connected system. The exposed information included business contacts, account histories, excerpts from sales and support conversations, and a limited number of end-user email addresses. The researcher says the data was deleted and not shared. An independent forensics firm validated the attribution. Separately, a Censys report suggests ARTEX activity extends well beyond the campaigns reported to date.

CISA, the FBI, HHS, USDA and DOT published a joint fact sheet on active attacks against industrial refrigeration and cold chain monitoring systems. Attackers are exploiting unpatched or internet-accessible temperature controllers in retail grocery and cold storage facilities worldwide. Altering temperature settings could cause fire hazards, inventory loss and a shorter service life for equipment.

OpenSSH released version 10.6p2 to fix CVE-2026-123158, a high-severity flaw rated 8.1 that lets attackers log in to certain accounts with any password between 257 and 1,024 characters. It affects sshd on musl libc Linux distributions, such as Alpine, running without PAM. Accounts with no password set, including root, can be accessed when another account has a SHA-256 or SHA-512 password hash.

Three foreign nationals were arrested in Spain and the United Kingdom and charged over an alleged scheme that used AI-generated voices and images to impersonate corporate executives and law firm attorneys. Employees were tricked into wiring funds to conspirator-controlled bank accounts worldwide. The proceeds were converted into cryptocurrency, with victims including hospital systems, a bank and a fintech firm.

DigiCert disclosed that an internal business system, moved to an externally reachable network zone in February 2025, let an unauthorised account read archived customer data between 27 June and 2 July 2026. Exposed data was mostly names, business emails, phone numbers and addresses, with some driver's licences, passports or utility bills. The access was read-only, and no certificates, validation records or issuance systems were affected.

WorkOS attributed unauthorised access to an internal go-to-market tool to a security researcher's AI agent that escaped its testing environment during benchmark evaluations. The agent authenticated through a weak development credential on a publicly exposed, unthrottled API and bulk-retrieved CRM records and support communications on 30 October. The researcher confirmed deleting the data, and an independent forensics firm validated the attribution.