F5 Rushes Out Fixes for Five Critical BIG-IP Crash-and-Hijack Bugs
F5 released out-of-band hotfixes for five critical stack overflows in BIG-IP APM and iRules that let unauthenticated attackers crash TMM and possibly run code, plus a high-severity APM SAML bug. Not exploited yet, and there's no workaround.
F5 has pushed out-of-band hotfixes for five critical BIG-IP flaws that an unauthenticated attacker can use to crash the traffic engine and, F5 says, possibly run code. BleepingComputer reports the bugs are stack buffer overflows in Access Policy Manager (APM) and in iRules, and that F5 offers no mitigations, only the fixes. None is known to be exploited yet, but F5 bugs have a habit of ending up there: CISA lists seven exploited F5 flaws, and an APM zero-day was used in attacks only in September.
For defenders
What’s affected: BIG-IP systems with APM or with particular iRules on a virtual server. F5 says vulnerable configurations let a remote, unauthenticated attacker terminate the Traffic Management Microkernel (TMM), disrupting traffic while it restarts, and “may perform RCE”. It’s a data plane issue; the management interface isn’t the route in.
| CVE | Component (F5) | CVSS v4 |
|---|---|---|
| CVE-2026-121805 | iRules (PEM::subscriber) | 9.3 |
| CVE-2026-120292 | iRules | 9.3 |
| CVE-2026-120285 | iRules | 9.3 |
| CVE-2026-116953 | iRules | 9.3 |
| CVE-2026-123637 | APM | 9.3 |
| CVE-2026-126822 | APM SAML (integer underflow, DoS) | 8.7 |
Status: not exploited per F5, no public exploit reported, none in CISA KEV. F5 released the hotfixes on Wednesday (K000164058, K000164239); each CVE’s K article lists the vulnerable configuration. Engineering hotfixes cover the high-severity APM SAML bug. Shadowserver counts more than 177,000 internet-exposed BIG-IP instances and over 14,600 with APM fingerprints, though it can’t say how many are patched.
What it means for you
BIG-IP boxes sit in front of a lot of company websites, VPN logins and apps, balancing the traffic and checking who gets in. You won’t have one at home. If your business runs one, or a hosting or IT provider does, the risk right now is an outage that a stranger can trigger from the internet, with a possible takeover behind it. Ask for the hotfixes to go in this week.
Do this now
- Check each BIG-IP against the configurations in F5’s K articles, starting with APM and any iRules using PEM::subscriber.
- Install the out-of-band hotfixes; there’s no workaround to lean on.
- Watch for unexpected TMM restarts, which would be the first visible sign of someone trying these bugs.
- Make sure the BIG-IP management interface isn’t reachable from the internet, even though these flaws don’t use it.
CVE at a glance
- CVSS
- 9.3 Critical (v4.0)
- EPSS
- 0.7% chance of exploitation in the next 30 days (higher than 54.8% of CVEs)
- Exploited in the wild
- Not on CISA’s KEV list
- Public exploit
- None reported in public exploit databases
- CVSS
- 9.3 Critical (v4.0)
- EPSS
- 0.7% chance of exploitation in the next 30 days (higher than 54.8% of CVEs)
- Exploited in the wild
- Not on CISA’s KEV list
- Public exploit
- None reported in public exploit databases
- CVSS
- 9.3 Critical (v4.0)
- EPSS
- 0.7% chance of exploitation in the next 30 days (higher than 54.8% of CVEs)
- Exploited in the wild
- Not on CISA’s KEV list
- Public exploit
- None reported in public exploit databases
- CVSS
- 9.3 Critical (v4.0)
- EPSS
- 0.7% chance of exploitation in the next 30 days (higher than 54.8% of CVEs)
- Exploited in the wild
- Not on CISA’s KEV list
- Public exploit
- None reported in public exploit databases
- CVSS
- 9.3 Critical (v4.0)
- EPSS
- 0.8% chance of exploitation in the next 30 days (higher than 56.5% of CVEs)
- Exploited in the wild
- Not on CISA’s KEV list
- Public exploit
- None reported in public exploit databases
- CVSS
- 8.7 High (v4.0)
- EPSS
- 0.5% chance of exploitation in the next 30 days (higher than 42.6% of CVEs)
- Exploited in the wild
- Not on CISA’s KEV list
- Public exploit
- None reported in public exploit databases
Data from NVD, FIRST EPSS and CISA KEV, refreshed automatically. We report whether public exploit code exists but never link to it: fake proof-of-concepts that infect researchers are common.