Daily Cybersecurity Brief

2026-06-07

Today: WordPress site takeovers are spreading via a critical Everest Forms Pro exploit that creates rogue admin accounts. Cisco SD-WAN systems face active zero-day attacks with no patch available. US fuel tank gauges are under attack, with 909 Internet-exposed devices concentrated stateside creating infrastructure disruption risks.

Critical Alerts

Cisco SD-WAN Zero-Day Under Active Attack

Cisco SD-WAN systems are being actively exploited through a zero-day vulnerability with no patch currently available. The vulnerability allows attackers to compromise SD-WAN infrastructure, potentially enabling network-wide access and traffic manipulation. Cisco has not released specific CVE details or severity scores yet.

• Action: Monitor Cisco security advisories for emergency patches. Implement compensating controls such as restricting SD-WAN management interface access to trusted networks only. Review SD-WAN logs for suspicious authentication attempts or configuration changes.
• Source: The Register

Critical Everest Forms Pro Flaw Exploited to Take Over WordPress Sites (CVE-2026-3300)

Hackers are exploiting CVE-2026-3300, a critical unauthenticated remote code execution vulnerability in Everest Forms Pro plugin versions 1.9.12 and earlier. The flaw exists in the Complex Calculation feature, where user input passed through forms is inserted into PHP code strings and executed via eval(). Attackers are injecting PHP code to create administrator accounts with username "diksimarina". Wordfence blocked over 29,300 exploitation attempts since active exploitation began April 13. EPSS score is 0.003 (55th percentile), indicating low predicted exploitation probability, but active exploitation is confirmed.

• Action: Update Everest Forms Pro to version 1.9.13 or later immediately. Review administrator accounts for suspicious usernames, especially "diksimarina". Check logs for strings matching "diksimarina" or form submissions with single quotes followed by PHP function calls. Block IP addresses 202.56.2.126 and 209.146.60.26 at firewall. If compromise is suspected, audit installed plugins, themes, and file system for webshells.
• Source: BleepingComputer

Exposed Fuel Tank Gauges Under Attack in the US

Threat actors are targeting Internet-exposed automatic tank gauge (ATG) systems at industrial facilities and gas stations across the United States. ATGs monitor liquid storage tanks for fuel, chemicals, and other materials. By compromising these systems, attackers can alter tank readings, disable safety alerts, or manipulate pump controls. CISA, FBI, NSA, DoE, EPA, TSA, DOT, and USDA issued a joint notice urging organizations to harden ATG systems. Shadowserver Foundation scans identified 909 exposed ATG devices in the US, representing over 90% of global exposure. Previous campaigns linked to Iranian threat actors have targeted gas station ATGs.

• Action: Immediately isolate ATG systems from direct Internet access. Implement network segmentation to separate ATG/SCADA systems from corporate networks. Review firewall rules to block inbound connections to ATG management interfaces. Enable logging and monitoring for ATG configuration changes and abnormal readings. Patch ATG firmware to latest versions where available. ATG devices often run legacy software and cannot run security tools, so network-level controls are critical.
• Sources: Dark Reading

Business & Infrastructure Threats

Adaptive AI Worms Loom as Next Enterprise Threat

Researchers at University of Toronto, Vector Institute, ServiceNow, Cambridge, and BeyondTrust have developed proof-of-concept agentic AI worms that autonomously propagate by searching for vulnerabilities, adapting to new environments, and creating exploit code dynamically. These worms use small, free AI models to power recursive reasoning loops that detect and exploit diverse vulnerabilities across systems. Unlike traditional worms that exploit a single bug, AI worms adapt in real-time and cannot be stopped by patching a specific vulnerability. BeyondTrust predicts AI-powered worm attacks targeting developers and engineers within six months to a year. Early examples like Shai-hulud (npm credential stealer) and GlassWorm (VS Code extension attack) have already combined self-propagation with credential theft.

• Action: Harden developer workstations and CI/CD pipelines as priority targets. Implement least-privilege access for service accounts and developer credentials. Monitor for unusual API calls to AI/LLM services from production systems. Deploy behavioral detection for lateral movement and automated exploitation patterns. Review and restrict outbound network access from development environments.
• Source: Dark Reading

ChatGPT Lockdown Mode Limits Data Exfiltration Tools

OpenAI has rolled out Lockdown Mode for ChatGPT personal accounts (Free, Go, Plus, Pro, and self-serve Business plans) to reduce data exfiltration risks from prompt injection attacks. Lockdown Mode disables live web browsing (restricts to cached content only), image display and retrieval, deep research, agent mode, Canvas networking, and file downloads. The feature does not prevent prompt injections but eliminates pathways for exfiltrating sensitive data to attacker-controlled infrastructure. Lockdown Mode and another unnamed feature cannot be enabled simultaneously. OpenAI also launched session management controls allowing users to review active sessions by device, location, and sign-in time.

• Action: Enable Lockdown Mode for accounts handling sensitive business data or proprietary information. Review active ChatGPT sessions and log out of unrecognized devices. Train users that Lockdown Mode does not prevent all prompt injection effects (malicious instructions in uploaded files can still affect behavior). Lockdown Mode is a mitigation layer, not a complete solution.
• Source: The Hacker News

Patch Priority

• Everest Forms Pro (WordPress): Critical unauthenticated RCE (CVE-2026-3300), active exploitation, update to 1.9.13+
• Cisco SD-WAN: Zero-day under attack, no patch available, restrict management access immediately
• Automatic Tank Gauges (ATG): Isolate from Internet, patch firmware, segment SCADA networks
• libexpat: Use-after-free in versions before 2.8.2 (CVE-2026-50219), update to 2.8.2+
• Python pip: Path traversal in console_scripts/gui_scripts (CVE-2026-8643), update pip to latest

Vulnerability Disclosures

CVE-2026-3300: Everest Forms Pro Unauthenticated RCE

Critical vulnerability in Everest Forms Pro for WordPress versions 1.9.12 and earlier allows unauthenticated remote code execution via PHP injection in Complex Calculation feature. EPSS 0.003 (55th percentile). Active exploitation since April 13 targeting admin account creation.

• Action: Update to version 1.9.13 or later. Audit admin accounts and logs for "diksimarina" username.
• Source: BleepingComputer

CVE-2026-50219: libexpat Use-After-Free Vulnerability

libexpat versions before 2.8.2 contain a use-after-free vulnerability due to missing handler call depth tracking when XML_GetBuffer, XML_Parse, XML_ParseBuffer, XML_ParserFree, or XML_ParserReset are called from within handlers during policy violations. EPSS 0.000 (2nd percentile).

• Action: Update libexpat to version 2.8.2 or later. Review applications using libexpat for XML parsing, particularly those handling untrusted XML input.
• Source: MSRC Security Update Guide

CVE-2026-8643: pip Path Traversal in Script Installation

Python package installer pip can extract console_scripts and gui_scripts outside the intended installation directory, allowing path traversal attacks. EPSS 0.000 (6th percentile).

• Action: Update pip to latest version. Review Python environments for unexpected files in system directories outside virtual environments.
• Source: MSRC Security Update Guide

CVE-2026-7774: Python tarfile Path Traversal Bypass

tarfile.data_filter in Python allows path traversal bypass, enabling malicious archives to write files outside the extraction directory. EPSS 0.000 (9th percentile).

• Action: Update Python to latest patched version. Review applications that extract tar archives from untrusted sources. Implement additional path validation before extraction.
• Source: MSRC Security Update Guide

CVE-2026-11332: Ansible-core Argument Injection in ansible-galaxy

Ansible-core vulnerable to argument injection in ansible-galaxy role install command, leading to arbitrary code execution. EPSS 0.000 (6th percentile).

• Action: Update ansible-core to latest version. Review ansible-galaxy role installation workflows, particularly in automated CI/CD pipelines. Validate role sources before installation.
• Source: MSRC Security Update Guide

CVE-2026-3276: Python DoS via Quadratic Complexity in unicodedata.normalize()

Python unicodedata.normalize() function vulnerable to denial-of-service through quadratic complexity when processing specially crafted Unicode input. EPSS 0.000 (16th percentile).

• Action: Update Python to latest patched version. Implement input length limits for user-supplied strings passed to unicodedata.normalize().
• Source: MSRC Security Update Guide

CVE-2026-43958: RRDtool Stack Buffer Overflow

RRDtool stack buffer overflow allows local code execution or denial-of-service. EPSS 0.000 (2nd percentile).

• Action: Update RRDtool to latest patched version. Review systems using RRDtool for monitoring and graphing, particularly those accepting input from untrusted users.
• Source: MSRC Security Update Guide

CVE-2026-10722: cilium eBPF Integer Overflow

cilium eBPF library vulnerable to integer overflow in LoadCollectionSpec/LoadCollectionSpecFromReader btf.go loadRawSpec function. EPSS 0.000 (2nd percentile).

• Action: Update cilium eBPF library to latest version. Review Kubernetes environments using cilium for network policy enforcement.
• Source: MSRC Security Update Guide

CVE-2026-37460: FRRouting BGP DoS Vulnerability

FRRouting (FRR) stable/10.0 through stable/10.6 vulnerable to denial-of-service via crafted BGP UPDATE messages due to missing input validation in rfapiRibBi2Ri() function. EPSS 0.000 (13th percentile).

• Action: Update FRRouting to version after stable/10.6. Review BGP peering configurations and implement BGP route filtering and validation.
• Source: MSRC Security Update Guide

CVE-2026-42504: Go mime Package Quadratic Complexity DoS

Go WordDecoder.DecodeHeader in mime package vulnerable to denial-of-service through quadratic complexity when processing malformed MIME headers. EPSS 0.000 (13th percentile).

• Action: Update Go to latest patched version. Review applications parsing email headers or MIME content from untrusted sources.
• Source: MSRC Security Update Guide

CVE-2026-42507: Go net/textproto Unescaped Input in Errors

Go net/textproto package includes arbitrary inputs in error messages without escaping, potentially enabling injection attacks when error messages are logged or displayed. EPSS 0.000 (9th percentile).

• Action: Update Go to latest patched version. Review error handling in applications using net/textproto to ensure error messages are sanitized before logging or display.
• Source: MSRC Security Update Guide

CVE-2026-27145: Go Inefficient Hostname Parsing in crypto/x509

Go crypto/x509 package vulnerable to resource exhaustion through inefficient candidate hostname parsing when validating certificates. EPSS 0.000 (1st percentile).

• Action: Update Go to latest patched version. Monitor TLS/certificate validation performance for unexpected slowdowns.
• Source: MSRC Security Update Guide

CVE-2026-8829: Perl HTML::Entities Use-After-Free

HTML::Entities versions before 3.84 for Perl read freed heap memory in _decode_entities function. EPSS 0.000 (9th percentile).

• Action: Update HTML::Entities to version 3.84 or later. Review Perl applications using HTML entity decoding, particularly web applications processing user input.
• Source: MSRC Security Update Guide

CVE-2026-5419: GnuTLS Timing Side-Channel in PKCS#7 Padding

GnuTLS vulnerable to information disclosure via timing side-channel in PKCS#7 padding removal. EPSS 0.000 (12th percentile).

• Action: Update GnuTLS to latest patched version. Review systems using GnuTLS for encrypted communications where timing attacks are feasible.
• Source: MSRC Security Update Guide

General Security News

Opal Security Raises $23 Million for AI-Native Identity Governance

Opal Security raised $23 million in funding led by Greylock and Battery Ventures, bringing total investment to $59 million. The company provides AI-native identity governance with real-time visibility and policy-as-code for employees, service accounts, and AI agents. Opal automates just-in-time access, revokes access based on risk and staleness, and enforces policies at machine speed. The company expanded headcount by 60% in 2026, primarily in engineering, product, and go-to-market roles.

• Source: SecurityWeek

Trends & Context

Today's threat landscape shows active exploitation of critical vulnerabilities in widely-deployed systems (WordPress plugins, Cisco SD-WAN, industrial controls), with attackers moving faster than patching cycles. The emergence of AI-powered adaptive worms represents a fundamental shift from single-exploit propagation to intelligent, multi-vector attacks that will require behavioral detection rather than signature-based defenses. Industrial control systems remain critically exposed, with US infrastructure disproportionately vulnerable due to Internet-facing management interfaces on legacy devices.