Four flaws, one government deadline
CISA added four vulnerabilities to its Known Exploited Vulnerabilities catalog on November 9: a GPUStack authentication bypass, a FreeSWITCH command-injection flaw, an OpenSSH authentication bypass affecting specific Linux configurations, and an IBM Aspera Faspex authorization bypass. All four have confirmed real-world exploitation, which is the bar CISA requires before a flaw earns a KEV listing in the first place, rather than sitting in the much larger pool of theoretically dangerous but unexploited CVEs.
Federal civilian agencies covered by Binding Operational Directive 26-04 had until November 12 to remediate all four and complete required forensic triage, a three-day window from the catalog addition that reflects how seriously the directive treats confirmed active exploitation. Private organizations face no such compliance deadline, yet matching its urgency is the right call given that attackers do not wait for a compliance clock to start their next move against unpatched systems.
The flaw that keeps giving attackers a way back in
GPUStack versions before 2.2.0 carry an authentication bypass that can hand a remote attacker full administrator access with no credentials at all, exploitable through certain reverse-proxy configurations or by spoofing the X-GPUStack-Real-IP header. The vendor has documented a real attack chain in which malicious inference backends were used to launch reverse shells, making this a confirmed technique already deployed against production systems rather than a theoretical proof-of-concept circulating in a research paper.
What makes this one genuinely dangerous beyond the initial compromise is persistence. An attacker who gained admin access before you patched may have created API keys or accounts that remain valid after the update ships, since the patch closes the entry point without retroactively invalidating anything created through it. Any organization running GPUStack needs to audit its account and API key list for anything unfamiliar, not just confirm the version number on the patch.
FreeSWITCH and OpenSSH: two very different exposure profiles
FreeSWITCH version 1.11.2 and earlier carries a command-injection flaw in SIP MESSAGE handling, rated a maximum-severity 9.8, that lets crafted messages reach an internal API command interface. From there, attackers can execute arbitrary commands, read configuration files and stored credentials, or manipulate active calls, which is a serious risk for any organization running voice infrastructure on the affected versions. The fix lands in version 1.11.3, and any unpatched instance exposed to the internet should be treated as already compromised until proven otherwise.
OpenSSH's issue is narrower but easy to misjudge. It affects certain Linux systems that use musl libc without PAM, a combination found in some Alpine Linux environments, and not every OpenSSH installation carries the exposure. That specificity cuts both ways: teams running the affected configuration face a genuine authentication bypass, but teams running a standard glibc-based distribution can likely rule themselves out quickly rather than burning triage time confirming a non-issue.
IBM's own numbers do not quite add up
IBM Aspera Faspex versions 5.0.0 through 5.0.19 carry an authorization bypass that can grant full administrative privileges, which IBM itself rates 9.8 on CVSS, its own maximum practical severity tier for a flaw of this kind. The fix ships in version 5.0.19 PL1, and given the active exploitation already confirmed, any organization running an affected version should treat this as a same-week patching priority rather than folding it into a routine maintenance cycle.
The detail worth flagging to your own security team is an inconsistency in IBM's bulletin: it describes an authenticated-user attack path while its published CVSS vector states that no privileges are required to exploit the flaw, and the bulletin does not reconcile that gap. When a vendor's own documentation disagrees with itself about prerequisites, the safe assumption for your risk assessment is the more permissive, more dangerous reading, not the more reassuring one.
Why patching is necessary but not sufficient
The pattern across these four flaws, especially GPUStack, is a reminder that a patch closes the door an attacker used to get in, but it does nothing about what they did once inside. Admin accounts created during a compromise window, API keys issued under false authority, and configuration changes made by an intruder all survive a version upgrade untouched unless someone goes looking for them specifically, which is a step most patch management workflows do not include by default.
Any system that was running a vulnerable version of GPUStack, FreeSWITCH, OpenSSH, or Aspera Faspex during the exploitation window deserves a credential audit and a log review in addition to the update itself. That is more work than checking a box on a patch compliance dashboard, but it is the difference between actually closing an incident and simply hiding the evidence of one that is still ongoing inside your environment.
What this means for your vulnerability management program
CISA's KEV catalog additions are a floor, not a ceiling, for how seriously to treat a flaw. The confirmation bar for inclusion is real-world exploitation, so by the time a vulnerability shows up on that list, the question is no longer whether attackers know about it, it is only how far they have already gotten into environments that have not yet patched, which should reframe how your team prioritizes the rest of this quarter's backlog.
If your organization runs any form of government contract or federal data pipeline, treat BOD 26-04's three-day remediation window as the realistic benchmark for your own response time on future KEV additions, not an aspirational target reserved for agencies alone. Attackers scanning for these specific flaws are not slowing down to check whether your organization happens to be a federal entity before they try their exploit against it.



