F5 has rushed out emergency patches for six critical flaws in its BIG-IP traffic-management appliances, warning that unauthenticated attackers can crash the devices that carry live web traffic — and may be able to run their own code on them.

The flaws, all stack-based buffer overflows rated 9.8 out of 10, were disclosed in out-of-band advisories on November 18. F5 says there is no workaround for any of them: installing the engineering hotfixes is the only fix.

Five of the six sit in iRules, the scripting feature administrators use to steer traffic through BIG-IP virtual servers. The sixth affects BIG-IP Access Policy Manager (APM), the module that handles remote-access and authentication policies. Each flaw needs a specific feature switched on before it can be exploited — but on a device with the trigger configured, no login is needed at all.

No login needed — and no workaround

An attacker who can reach a vulnerable virtual server can crash the Traffic Management Microkernel (TMM), the process that moves every client connection through the device. Traffic stops until TMM restarts, so even a denial-of-service outcome takes websites, remote-access portals and dependent applications offline. F5 says the same flaws may also allow remote code execution.

The exposure is in the data plane — the part of the device that handles live traffic — not the management interface. Locking down the admin console does not help; what matters is whether the affected iRule commands or APM policy are configured on an internet-facing virtual server.

A seventh flaw disclosed in the same batch, CVE-2026-126822, is a high-severity integer underflow in APM's SAML handling that lets an unauthenticated attacker trigger a denial of service. It is rated 8.7.

More than 177,000 devices exposed

Internet-scanning project Shadowserver counts more than 177,000 internet-exposed BIG-IP instances, including over 14,600 carrying APM fingerprints. F5 has not reported any exploitation of this batch, and the flaws were not in CISA's Known Exploited Vulnerabilities catalog when vendors checked on November 18. But history argues against waiting: CISA has tagged seven F5 vulnerabilities as actively exploited since November 2021, and in September F5 patched a critical APM zero-day, CVE-2026-94127, that was already under active attack.

The affected lines are BIG-IP 17.1.0 to 17.1.3, 17.5.0 to 17.5.1 and 21.1.0, plus BIG-IP Next for Kubernetes and Cloud-Native Network Functions. The correct hotfix depends on the exact build installed. BIG-IQ, F5OS, F5 Distributed Cloud, NGINX products and F5 AI Gateway are not affected by these six flaws — a different set of NGINX flaws was patched separately.

What to patch first

Security teams should start with internet-facing virtual servers running the trigger configurations: iRules using PEM::subscriber, MR::message route, PSC::ip_address, GENERICMESSAGE::route or DNS::scrape, and APM per-request policies that combine the Client Information and Endpoint Security agents. F5 advises patching the standby unit of each high-availability pair first, failing over, and then patching the former active unit.

Until the hotfix is installed, the only options are to detach or rewrite the matching iRule, or restrict which source addresses can reach the affected virtual server. F5 lists no mitigation of its own, and removing an APM posture check weakens a security control — neither is a substitute for patching. Administrators should also watch for unexplained TMM restarts: a crash is worth treating as an incident until its cause is known, because the possible impact includes code execution.