ZeroFox Daily Intelligence Brief - October 23, 2026

ZeroFox Daily Intelligence Brief - October 23, 2026

ZeroFox Intelligence collects, curates, and analyzes information derived from open and proprietary sources. Here is today’s daily roundup to give you and your clients an advantage over the adversary.

Brief Highlights

• ZeroFox Intelligence Flash Report - AI-Discovered PHP Vulnerability
• Developer Cloud Platform Railway Reports Potential Credential Exposure
• Hackers Compromised LiteLLM’s Release Pipeline to Steal Credentials

ZeroFox Intelligence Flash Report - AI-Discovered PHP Vulnerability

Source: https://www.zerofox.com/advisories/42749/

What we know: Exploitation attempts of the recently-patched PHP2Shell vulnerabilities have reportedly surged, while thousands of attempts were reportedly made in the hours before public announcement, rising to millions within two hours of disclosure. The PHP Foundation released emergency patches on October 21, 2026 for two critical vulnerabilities, collectively dubbed “PHP2Shell” that enable unauthenticated remote code execution on unpatched servers.

Context: Both flaws were discovered using frontier AI models in August 2026, with AI also reducing the effort required to build working exploits. The root cause has been present in the PHP codebase for 14 years. An initial industry analysis identified 22.6 million potentially vulnerable instances. PHP has released patches for all active branches (8.2 through 8.5), while legacy versions—including 8.1 and earlier—have not received official patches.

Analyst note: ZeroFox assesses that mass exploitation of unpatched PHP servers will almost certainly continue through at least the coming weeks. Initial activity is likely to consist of opportunistic scanning, web shell deployment, and credential theft, with initial access sales and follow-on ransomware activity likely to follow. PHP2Shell will very likely accelerate the use of AI by both defenders and threat actors to find and weaponize flaws in long-established codebases.

Developer Cloud Platform Railway Reports Potential Credential Exposure

Source: https://thehackernews.com/2026/10/railway-says-intruder-may-have-accessed.html

What we know: Cloud hosting platform Railway suspects that an unauthorized third party "may" have accessed systems storing customer environment variables and obtained their decrypted contents. The exposed variables can contain sensitive information, including API keys, database credentials, and private registry credentials.

Context: Security researchers reportedly claim that attackers used cryptocurrency private keys stored on Railway to drain wallets and that the first identified theft occurred roughly two hours before Railway publicly disclosed the incident. Additionally, Railway believes the incident is likely part of a series of attacks against platform companies ultimately intended to compromise AI inference providers.

Analyst note: Further unauthorized activity is likely if exposed credentials remain valid, particularly where customers have not yet rotated secrets stored on Railway. Depending on the permissions attached to individual credentials, attackers are likely to access connected cloud services, databases, or AI services.

Hackers Compromised LiteLLM’s Release Pipeline to Steal Credentials

Source: https://www.bleepingcomputer.com/news/security/litellm-release-pipeline-hacked-to-distribute-info-stealer-malware/

What we know: Hackers have compromised the LiteLLM project’s PyPI package and published malicious Docker images containing malware designed to steal credentials and load remote-code execution tools.

Context: Malicious Python startup files caused the malware to execute automatically whenever Python launched in an affected environment. The attack affected all seven PyPI installation packages for LiteLLM version 1.107.0, along with multiple Docker Hub and GitHub Container Registry images, potentially impacting Windows, Linux, and macOS systems. The malware targeted cloud and deployment credentials, AI credentials and configuration, developer secrets and private keys, browser data, and host and runtime information. The malicious artifacts have been removed, and version 1.106.2 is the recommended clean release.

Analyst note: Stolen credentials could enable hackers to access cloud infrastructure, AI services, developer accounts, and CI/CD pipelines, potentially leading to further data theft, system compromise, and malware deployment

DEEP AND DARK WEB INTELLIGENCE

Exploit user ErnestoCheGuevar: Untested threat actor “ErnestoCheGuevar” has advertised a dataset allegedly associated with organizations in the European Union (EU) and the United States on a predominantly Russian-language dark web forum Exploit. The dataset allegedly contains personally identifiable information (PII) of a large number of individuals registered on state registries, airlines, large retail firms, and boutiques. The actor did not provide a sample or other details regarding the dataset. The dataset is likely to be recycled records from earlier data breaches given the lack of details and sample data in the post.

VULNERABILITY AND EXPLOIT INTELLIGENCE

CVE-2026-114625: Arista Networks has released security updates for this actively exploited zero-day vulnerability affecting VeloCloud Orchestrator (VCO) On-Prem deployments. The flaw can enable authenticated attackers with basic privileges to escalate their access, impersonate a superuser operator, and compromise the confidentiality, integrity, and availability of the VeloCloud Orchestrator host and its managed data.

Affected products: The affected products are listed in this advisory.