On this page (7 sections)
A critical information disclosure flaw, tracked as CVE-2026-123551, affects OpenAM. According to the NVD record, unauthenticated attackers can read live session tokens from OpenAM's notification WebSocket and use them to hijack user and administrator sessions. The CVSS score is 10.
If you run OpenAM as your single sign-on or access management layer, anything it protects is exposed when its sessions can be stolen. Teams running versions from 14.0.0 up to (but not including) 16.1.4 should treat this as urgent.
What happened
The NVD entry, published on 2026-11-08, describes a flaw in how OpenAM protects its notification WebSocket. That channel is meant to be used only by agents, the components that talk to the OpenAM server. An unauthenticated outsider is able to get past that restriction and read the live session tokens it carries.
The tokens named in the record are iPlanetDirectoryPro tokens, the cookie OpenAM uses to represent an authenticated session. Anyone holding a valid one can generally act as that user, including administrators, until the session ends.
Technical details
The weakness is classed as CWE-201, which covers sensitive data leaking through a message that the system sends. Here the sensitive data is session tokens, and the leak is on a WebSocket meant for trusted components.
According to the record, the agent-only check can be bypassed through an unvalidated internal forward on legacy Liberty ID-FF endpoints. In plain terms, a request to an old federation endpoint is passed along inside the server without the check that would normally stop a non-agent, and it then reaches the WebSocket as if it were legitimate. Liberty ID-FF is an older federation protocol, so deployments that never use it may still have those endpoints present.
The research does not give the exact request paths, whether a particular configuration reduces exposure, or any proof-of-concept. Those details are unconfirmed, and this article does not speculate on them.
Who is affected
- Organizations running OpenAM from 14.0.0 up to but not including 16.1.4.
- Deployments where the OpenAM server is reachable from the internet or from a large internal network, since the attacker needs no credentials.
- Environments where administrators sign in through OpenAM, because a stolen administrator token can give control over the access management system itself.
- Applications that trust OpenAM sessions for single sign-on, since a hijacked session may work against the applications that rely on it.
The research notes that only one independent publisher covers this story so far. The main technical references are the NVD record and the advisory pages linked from it, including the project's GitHub security advisory. Details may be refined as the vendor and others publish more.
How to check your exposure
Start by finding out which OpenAM version you run, for example from your deployment records or the build you installed. If you cannot confirm it, assume you are in the affected range until proven otherwise.
Then check whether the server is reachable from places it does not need to be, and review your logs. The command below is read-only. It assumes an nginx access log at the default Debian/Ubuntu path and lists the last 100 requests that carry a WebSocket upgrade header, which only works if your log format records that header. Adjust the path and format to your environment, and look for source addresses you do not recognize.
grep -i 'websocket' /var/log/nginx/access.log | tail -n 100This is only a starting point. The research does not state which request paths the attack uses, so no specific path pattern can be recommended, and a clean result does not prove nothing happened. Default access logs often do not record upgrade headers at all, and proxy setups vary.
What to do now
- Identify every OpenAM instance you run, including test and forgotten ones, and record its version.
- Upgrade any instance from 14.0.0 up to 16.1.3 to 16.1.4 or later, following the vendor's guidance in the project's GitHub security advisory.
- Until you can patch, restrict network access to the OpenAM server so only the clients and components that need it can reach it.
- If you do not use Liberty ID-FF, ask the project whether those legacy endpoints can be disabled. The research does not confirm a supported workaround.
- After patching, invalidate existing sessions where your deployment allows it, so any token already collected stops working. Have administrators sign in again.
- Review administrator accounts, recent configuration changes and newly created users or policies for anything you did not make.
- If you see any sign of suspicious access, rotate secrets that an attacker with an administrator session could have read.
Detection and monitoring
Because the flaw is about stolen but otherwise valid sessions, the signs may look like normal logins. Watch for the same session token used from two different IP addresses or locations, administrator actions at unusual hours, and new administrator or policy changes with no matching change ticket.
Keep alerts on OpenAM's authentication and audit logs, and forward them somewhere an attacker with an administrator session cannot easily edit. It also helps to monitor the public side of your login infrastructure. Webforta Monitor can watch uptime, certificate and DNS changes on the hostnames in front of your identity system, which helps you notice unexpected changes around it, though it does not detect or block this specific flaw.
For general guidance on responding if you suspect compromise, see website hacked: what to do. A web application firewall can add a layer in front of exposed services; what is a WAF explains what it can and cannot do.
Timeline
| Date | Event |
|---|---|
| 2026-11-08 | CVE-2026-123551 published in the NVD with a CVSS score of 10 |
Sources
This article is based on the public sources below. Details can change as vendors and researchers publish more; check the original advisories before acting on version numbers.
