Rockwell Automation’s FactoryTalk Analytics LogixAI contains a high-severity configuration weakness that could expose sensitive operational data to attackers on adjacent networks. The U.S. Cybersecurity and Infrastructure Security Agency (CISA) republished an advisory assigning CVE-2025-9364 to an overly permissive Redis instance used by the industrial analytics module, with CVSS scores hovering around 8.8. The vulnerability affects versions 3.00 and 3.01, and operators across critical manufacturing sectors worldwide are being told to upgrade to version 3.02 or later immediately.

FactoryTalk Analytics LogixAI, often deployed as an edge machine-learning engine alongside Rockwell’s ControlLogix systems, ingests controller tags to run predictive models and soft sensors. The product is marketed as a no-code analytics tool for process anomaly detection, but CISA’s advisory reveals that its embedded Redis database—if left with default or weak configurations—can be reached and manipulated by anyone with network access to the internal plant floor.

A Redis Misconfiguration, Not a Code Bug

The vulnerability falls under CWE-497: Exposure of Sensitive System Information to an Unauthorized Control Sphere. In plain language, the LogixAI versions in question ship or default to a Redis setup that listens on non-loopback interfaces, lacks authentication, and does not enforce access controls. Redis, an in-memory data store popular for caching and real-time analytics, is notorious for being exposed when basic hardening steps are skipped. Here, the advisory points to an “over permissive Redis instance”—meaning it is bound to a broad network address (possibly 0.0.0.0), requires no password, and allows administrative commands that can read and alter stored data.

An attacker on the same operational technology (OT) LAN, or on a poorly segmented business network that can reach the OT zone, could connect to the default Redis port (TCP 6379) and issue commands to dump analytics artifacts, trained models, or even modify prediction baselines. The attack complexity is low; no authentication bypass or memory corruption is needed—just a network path and a Redis client.

Operational Impact: More Than Data Theft

Because LogixAI sits close to process control, the consequences extend beyond simple information disclosure. Sensitive process parameters, product recipes, and real-time sensor streams could be stolen, enabling industrial espionage or insider threats. Worse, an adversary who tampers with model parameters or prediction outputs could generate false alarms or, more dangerously, false assurances that mask actual process anomalies. Operators relying on LogixAI’s recommendations might misjudge equipment health or product quality, leading to safety incidents or production losses.

The Redis instance also becomes a pivot point. Once inside, an attacker could enumerate the connected controllers, harvest credentials from Redis configuration files, or use the analytics host as a beachhead to move laterally into deeper OT systems. CISA’s advisory underscores the global deployment of LogixAI and its classification within critical manufacturing—food, chemicals, energy, and other sectors where uptime and integrity are paramount.

Immediate Mitigation Steps

Rockwell has released version 3.02 to address the misconfiguration, and CISA’s recommended path is unambiguous:

  • Inventory and patch: Identify every LogixAI instance running 3.00 or 3.01. Upgrade to 3.02 or later after staging tests.
  • Network segmentation: Isolate LogixAI hosts on dedicated OT management networks. Block inbound TCP/6379 from all but explicitly trusted admin hosts. Use firewalls, ACLs, and microsegmentation to prevent lateral access.
  • Redis hardening (if patching must be delayed): Bind Redis to localhost or a specific management interface. Enable ACLs or requirepass authentication. Disable or rename dangerous commands like CONFIG, FLUSHALL, DEBUG, and SHUTDOWN, though this must be done consistently across any replicas to avoid replication breaks.
  • Enable TLS: Encrypt Redis traffic to prevent credential sniffing on the wire.
  • Monitor aggressively: Log unexpected Redis connections, watch for anomalous commands, and alert on any sudden changes to model artifacts or prediction error rates.

Detection and Incident Response

Operators should integrate Redis monitoring into their SIEM or OT-specific detection tools. Indicators of potential compromise include remote client connections from unknown IPs to TCP 6379, execution of privileged commands, or unexplained shifts in model timestamps. If exploitation is suspected, isolate the LogixAI host immediately while preserving volatile data, snapshot the Redis database, and compare model integrity against offline backups. Only then should the 3.02 upgrade be applied in a controlled manner.

A Recurring Industrial Pattern

This advisory is the latest in a line of Rockwell and ICS-focused alerts that reveal misconfigurations rather than zero-day exploits. Modern OT stacks increasingly bundle open-source components—Redis, Node.js, web frameworks—and security often hinges on proper setup, not just patch management. CISA has repeatedly warned that network exposure combined with default credentials remains a top risk for industrial environments. The LogixAI case reinforces the need for asset owners to treat embedded services as part of the attack surface, hardening them even if the vendor documentation doesn’t emphasize it.

What the Fix Covers and What It Doesn’t

Upgrading to LogixAI 3.02 should reconfigure the Redis instance with safer defaults—binding to localhost, enabling authentication, or removing unnecessary exposure. However, the advisory does not detail the exact change, and operators who have customized their deployments should verify the Redis binding and ACL settings after the update. Moreover, if the upgrade is applied in a complex high-availability topology with Redis replication, additional steps may be needed to ensure that security controls are consistent across all nodes.

Timeline for Remediation

  • Days 0–7: Complete an inventory of all LogixAI assets. If version 3.00 or 3.01 is found, apply network-level restrictions immediately (firewall blocks, host-based firewalls).
  • Weeks 1–4: Stage the upgrade to 3.02 in a test environment, validate model outputs, then roll out to production. Implement full Redis hardening (ACLs, TLS, monitoring).
  • Months 1–3: Audit other OT appliances and edge devices for similar embedded services. Expand SBOM and dependency checking into procurement processes to catch misconfigurations before they reach the plant floor.

CISA’s advisory (ICSA-25-252-08) remains the authoritative source, and operators should consult Rockwell’s compatibility and download center for the corrected build. Redis’s official security documentation provides the hardening fundamentals that would have prevented this exposure. With low attack complexity and a clear fix, CVE-2025-9364 is a problem that can be solved with network discipline and patching—but only if organizations act before adversaries capitalize on exposed analytics engines.