Broadcom released emergency patches for five VMware flaws, three of them critical, affecting ESXi, vCenter, Workstation, and Fusion. The most notable, CVE-2026-47876, is an out-of-bounds write in the ESXi VMXNET3 network adapter that lets an attacker with admin rights inside a guest virtual machine run code on the host, a VM escape. Two critical vCenter flaws follow: CVE-2026-59309 is an authentication bypass reachable over the network, and CVE-2026-59310, scored 9.8, allows code execution through a directory traversal. Broadcom warns the vCenter bypass could be chained with the escape or the code-execution flaw to seize the hypervisor without any guest foothold. No exploitation is reported yet.
The Ruby on Rails team disclosed a critical flaw in Active Storage that lets an unauthenticated attacker read arbitrary files by uploading a crafted image. Tracked as CVE-2026-66066 and scored 9.5, it affects applications that use the libvips image library and accept image uploads from untrusted users, which is the default in modern Rails. Active Storage passes uploads to libvips without disabling its unsafe image loaders, so a malicious file can read the server process environment, exposing the secret key base, database passwords, cloud storage keys, and API tokens. Those secrets can enable code execution and lateral movement. Rails is not aware of exploitation, and full details are held until August 28.
Noma Labs found a critical flaw in Ruflo, an open-source platform for orchestrating swarms of AI agents that acts as a harness for coding agents like Claude Code and Codex. Tracked as CVE-2026-59726 and scored 10.0, the flaw is that Ruflo's default Docker deployment exposed its Model Context Protocol bridge to the network with no authentication, offering 233 callable tools. A single unauthenticated request could invoke a shell command tool to run code in the bridge container, read AI provider API keys, steal stored conversations, and poison the platform's persistent agent memory, which can steer future outputs even after the attacker leaves. The flaw is in Ruflo, not in the coding agents.
A breach at SplitVPN, a service formerly called NotVPN that marketed itself as keeping no logs, exposed a 17GB database containing roughly 58 million connection logs. The logs record which device connected to which server and when, running continuously up to the day of the breach, directly contradicting the no-logs promise. Cross-referenced with user and device tables holding emails, last-seen IP addresses, and hardware identifiers, they can reconstruct who connected from where and when for tens of millions of people. The data also includes about 23 million user records and 2.6 million payment records with masked card details. Operator account hashes were exposed too.
Attackers published malicious versions of the @joyfill/components and @joyfill/layouts npm packages that run a remote access trojan as soon as the package is imported, not merely installed. Because the payload executes at import time, defenses that block install scripts, such as installing with scripts disabled, do not stop it. The malicious code sat only in the published tarballs with no matching source change, pointing to a registry or publishing pipeline compromise. Once loaded, it profiles the host, opens a remote-control channel, and can run shell commands, upload files, read the clipboard, and tamper with developer tools, using blockchain networks to resolve its next payload for resilience.
New research shows that application security scanners embedded in the software pipeline can themselves be attacked, because analyzing a repository is not always a read-only action. Many scanners execute code or process files in ways that can be forced to run attacker content when it is handled without isolation. For example, a tool that scans cloud configuration files against a folder of custom rules can be fed a rule that is actually a malicious program, which the scanner runs when it loads it. Since scanners often hold broad access to source, secrets, and build systems, a compromised one is a strong pivot point into the wider environment.
Arista patched a critical vulnerability in on-premises VeloCloud Orchestrator, the console that configures and manages VeloCloud SD-WAN deployments, that is already being exploited. Tracked as CVE-2026-16812 and scored 10.0, it is an unauthenticated operating-system command injection that lets a remote attacker reach privileged functionality meant only for internal use. Arista warns successful exploitation can compromise the confidentiality, integrity, and availability of the orchestrator and everything it manages. There is no configuration that prevents exposure, since the web interface is reachable by default. CISA added it to its exploited-vulnerabilities catalog. Hosted and Dedicated deployments were fixed before the advisory and are not affected.
n8n patched a high-severity flaw that lets an authenticated user who can create or edit workflows escape the expression sandbox and run operating-system commands as the n8n process. Security Joes found it while probing n8n's February fix for an earlier sandbox bug, and the gap sits in how the rewriter handled a bare identifier in a concise arrow function body. Successful exploitation can expose the n8n encryption key and allow decryption of stored credentials. It is tracked as GHSA-gv7g-jm28-cr3m with a score of 8.7 and no CVE assigned yet. Fixed versions are 2.31.5 and 2.32.1, with no patched 1.x release listed.
Public exploit details released on July 27 show how an unauthenticated request can reach PHP's eval function inside vBulletin and run code on an unpatched forum server. The attack needs no account, administrative access, or interaction from another user, making any internet-facing vulnerable forum a direct target. SSD Secure Disclosure lists vBulletin 6.2.1 and earlier and 6.1.6 and earlier as affected, without giving a lower version boundary. Forum software is an attractive target because it is public by design, often runs with limited maintenance, and a single compromised server can host web shells, redirect visitors, or expose community member data. Administrators should confirm their version and patch level.
Researchers showed that confused deputy weaknesses persist in Google Cloud and Microsoft Azure, where a trusted service can be tricked into acting on an attacker's behalf against resources it should not reach. The pattern shows up when one service holds broad permissions and accepts instructions or identifiers from a less-trusted source without verifying who is really asking, which can enable cross-tenant access or privilege escalation. It is the same class of issue behind recent findings in enterprise agent and integration tooling, where a component with a user's permissions follows attacker-supplied input. The takeaway is architectural: identity and authorization must be checked at every trust boundary, not assumed from the calling service.
TrustStrike Labs