CISA Orders Federal Agencies to Patch Actively Exploited F5 BIG-IP Flaw
A bug initially dismissed as a denial-of-service flaw is now an actively exploited remote code execution threat, with federal agencies given just days to patch.

What appeared to be a routine denial-of-service bug in F5 Networks' enterprise networking gear turned out to be something far more dangerous: a fully exploitable remote code execution flaw that attackers were already weaponizing before federal officials could raise the alarm.
CISA added CVE-2025-53521, a critical vulnerability in F5's BIG-IP Access Policy Manager, to its Known Exploited Vulnerabilities catalog on Friday, ordering federal civilian agencies to assess and remediate the flaw by Monday, March 30. The two-day window reflects the urgency of a threat that escalated faster than its original classification suggested.
F5 Networks initially categorized the flaw as a denial-of-service vulnerability. New information that emerged in March 2026 prompted the company to reclassify it as a remote code execution bug, revising its severity scores to 9.8 under the CVSS v3.1 framework and 9.3 under CVSS v4.0, placing it at the top of the severity spectrum. The flaw targets the apmd process, the component responsible for processing live traffic, in BIG-IP APM versions 17.5.0 through 17.5.1, 17.1.0 through 17.1.2, 16.1.0 through 16.1.6, and 15.1.0 through 15.1.10. Systems running in Appliance mode are also vulnerable.
"When a BIG-IP APM access policy is configured on a virtual server, specific malicious traffic can lead to remote code execution," F5 stated in its updated advisory.
The technical trigger is precise but not difficult to satisfy: an attacker needs only to send specially crafted traffic to a BIG-IP device where an APM access policy is active on a virtual server. No authentication is required. The result is full device compromise, meaning an attacker can pivot deeper into a protected network, steal sensitive data, or deploy ransomware across every system behind the gateway.
Security researchers tracking active exploitation have documented techniques designed to outlast simple detection. Observed attacker behavior includes the deployment of memory-resident web shells that leave no trace on disk and the use of disguised network traffic crafted to evade signature-based defenses. F5 separately confirmed that at least one threat actor modified components of sys-eicheck, the BIG-IP system integrity checker, as part of a post-compromise effort to obscure the intrusion.

The stakes extend well beyond federal networks. BIG-IP appliances sit at critical chokepoints for thousands of organizations: hospitals managing patient record systems, state and local governments routing public services, school districts supporting remote users, and financial institutions guarding transaction infrastructure. When a device serving as a load balancer or remote-access concentrator is compromised at that level, every application and user behind it is exposed.
CISA's KEV mechanism, established under Binding Operational Directive 22-01, carries legal weight for federal civilian executive branch agencies but functions as a strong public warning for the private sector. The catalog entry for CVE-2025-53521 signals that exploitation is not theoretical: incident responders are already finding it in production environments.
Organizations running any of the affected BIG-IP APM versions should treat anomalous activity in existing logs as potentially genuine rather than noise. The first step is inventorying every internet-facing BIG-IP device and confirming whether APM access policies are active on virtual servers. F5's knowledge base entry K000156741 provides patch and mitigation guidance; where immediate patching is not possible, disabling the APM access policy on affected virtual servers is the recommended interim control.
The rapid reclassification of CVE-2025-53521 from a low-priority disruption bug to an actively exploited RCE illustrates a persistent pattern: severity scores assigned at initial disclosure frequently underestimate what determined attackers can extract from a flaw once they invest time to study it. By the time the score is corrected, the exploitation window is already open.
Sources:
Know something we missed? Have a correction or additional information?
Submit a Tip

