IBM App Connect Enterprise CVE-2026-17156

Two concrete prerequisites must both hold: first, the attacker needs a local presence on the host running IBM App Connect Enterprise - a local account, shell, or the ability to write files into a directory that an integration node, integration server, or the Toolkit reads; second, the crafted serialized object must actually be deserialized, which the CVSS UI:R metric indicates requires a user or automated action such as importing, loading, or restarting with a manipulated artifact. Exploitation is limited by these factors: there is no network attack vector (AV:L), so a remote attacker cannot reach the flaw directly and would first have to obtain local access or a file-write primitive; the affected deserialization routine is only reachable in configurations where serialized input from an attacker-influenceable location is processed, so hardened deployments that lock down import paths and configuration directories are materially less exposed; and the PR:N metric suggests no elevated OS privilege is needed beforehand, though that should be verified against the vendor's description since the flaw appears to involve an operator-facing workflow rather than an unauthenticated interface.

The signals do not point in the same direction, and defenders should read them carefully. CVSS 7.8 with S:U/C:H/I:H/A:H says that if the flaw is reached, the impact on that host is complete, but the vector is AV:L/AC:L/PR:N/UI:R, meaning the attacker must already have a foothold on or access to the local machine and must additionally persuade a user or scheduled action to process the crafted payload - this is a privilege-escalation/code-execution primitive, not a remotely reachable pre-auth bug. SSVC agrees: Exploitation 'none', Automatable 'no', Technical Impact 'total' - a fix-when-convenient-to-out-of-cycle posture rather than emergency patching. There is no CISA KEV listing and no public exploit code identified at time of analysis, so no confirmed active exploitation. Note one internal tension to verify with the vendor: PR:N alongside AV:L is unusual and may reflect a default-configuration local process rather than a truly unprivileged local user; if the vulnerable deserialization point is actually only reachable by an authenticated operator, effective privileges are higher than PR:N implies. Even so, a mid-sized municipal IT team in olsztyn running App Connect for integration workloads should treat this as urgent for internet-facing or multi-tenant integration nodes, where a low-privileged local user escalating to the service account would be severe. EPSS data was not supplied, and no proof-of-concept status is stated - treat the absence of exploit code as unconfirmed rather than as a guarantee.

An attacker with an existing low-privileged account on the App Connect Enterprise host (or write access to a directory the server reads) crafts a serialized Java object containing a malicious gadget chain and places it where the product will deserialize it - for example a configuration or import artifact processed during a user-initiated operation. When a legitimate operator or scheduled task loads that artifact, deserialization triggers and the attacker's code executes with the privileges of the App Connect Enterprise service account. No public exploit code has been identified, so this currently requires custom tooling rather than a ready-made payload.