What happened
Oracle has disclosed two serious vulnerabilities in Oracle Access Manager, part of the Oracle Fusion Middleware suite. Both sit in its Authentication Engine, the component that decides who may log in and what they may access. The flaws are tracked as CVE-2026-91510 and CVE-2026-91507.
Records for both flaws appeared in the US National Vulnerability Database (NVD) on 17 November 2026. They point to Oracle’s security bulletin (cspunov2026), where the vendor publishes details and patches.
Is it being exploited? As of 18 November 2026, neither flaw is listed in the CISA KEV catalog, and the EU vulnerability database EUVD (run by ENISA) does not record any exploitation. Oracle’s description says nothing about attacks. That can change quickly, because Oracle rates both flaws as easy to exploit.
Why it is dangerous
Oracle Access Manager works like a gatekeeper for company applications. A user logs in once, and the gatekeeper then tells every other system who they are and where they may go. If someone tricks the gatekeeper or rewrites its lists, they open doors to everything behind it.
According to Oracle’s description, both flaws are easy to exploit and require no login and no action from a user. The attacker only needs network access to the server over HTTP. If the login page is reachable from the internet, which is common for single sign-on, the attack can come from anywhere. Oracle also flags a scope change: an attack may significantly affect other products, not just Access Manager itself.
- CVE-2026-91510 (CVSS 10.0): an attacker can create, delete or modify critical data or all data accessible to Access Manager, gain full read access to it and partially disrupt the service.
- CVE-2026-91507 (CVSS 9.3): an attacker can create, delete or modify critical data or all data accessible to Access Manager and read a subset of it.
The description does not mention running arbitrary code on the server. Still, taking control of the system that governs access across the whole organisation is serious in itself.
CVSS is a severity scale from 0 to 10. Both scores were assigned by Oracle itself using CVSS 3.1, and 10.0 is the highest possible value. The CISA KEV catalog lists flaws that the US agency CISA has confirmed as exploited, and these two are not on it yet.
EPSS estimates the probability that a flaw will be exploited in the next 30 days. As of 18 November 2026 it stands at 0.46% for CVE-2026-91510 and 0.34% for CVE-2026-91507. That is low for now, which is typical for freshly published flaws.
Who is affected
The affected supported versions are Oracle Access Manager 12.2.1.4.0 and 14.1.2.1.0, for both flaws.
Oracle Access Manager is used mainly by larger companies, public institutions and organisations running Oracle environments, to provide single sign-on (SSO) to intranets, portals, ERP and other web applications. It usually runs on the Oracle WebLogic application server and is often reachable from the internet so that staff and partners can log in from outside. It is typically managed by a supplier or integrator.
If you do not use Oracle Access Manager, these flaws do not affect you. The records list only this product. Oracle only assesses supported versions, so if you run an older, unsupported release, do not assume it is safe.
How to check if you are affected
First, find out whether you run Oracle Access Manager at all. Ask your supplier, or check whether an Oracle login page appears when you sign in to company applications. On the server, the opatch tool shows the version and installed patches. The lsinventory and lspatches commands below only read data and change nothing. Run them as the user that owns the Oracle installation.
$ORACLE_HOME/OPatch/opatch lsinventory
$ORACLE_HOME/OPatch/opatch lspatches
In the output, find the Oracle Access Manager version and compare the installed patches with the numbers in Oracle’s bulletin. Also check whether Access Manager addresses are reachable from the internet, for example in your firewall or reverse proxy rules.
How to fix it
The NVD records do not name specific fixed versions or patch numbers. You will find them in Oracle’s security bulletin at https://www.oracle.com/security-alerts/cspunov2026.html, and the patches are downloaded through My Oracle Support. Oracle’s description of the flaws does not mention a workaround. Because the flaws are not in the CISA KEV catalog, CISA has not issued a required action for them yet.
- Identify which servers run Access Manager and which versions you have.
- If possible, restrict internet access to Access Manager right away, for example via VPN only or from known IP addresses. This is not a fix, it only lowers the risk until the patch is in place.
- Back up the Oracle Home, the WebLogic domain configuration and the Access Manager configuration.
- Download the patches for your version from Oracle’s bulletin and follow their README. If you have a test environment, try them there first.
- Apply the patches in production, restart the servers and confirm with
opatch lspatchesthat they are installed. - If you run an unsupported version, plan a move to a supported one, as patches may not exist for older releases.
Order: internet-facing servers first, internal ones second. If a supplier manages Access Manager for you, agree on a patching date with them today.
What if you may already be compromised
Oracle’s description of the flaws does not include specific indicators of compromise. Look for unusual activity in general over the past few weeks.
- Access Manager and WebLogic server logs: unusual request volumes, authentication errors, requests from unknown IP addresses.
- Unplanned changes to the Access Manager configuration: new or modified policies, authentication schemes, protected applications or agents.
- New or changed accounts and groups in the directory used by Access Manager, especially with administrator rights.
- Logins to protected applications that do not match normal user behaviour, for example at night or from abroad.
Since an attack may reach the systems behind Access Manager, check their logs too. If you find anything suspicious, do not delete logs or reinstall the server before evidence is secured. Bring in an incident response specialist to determine the scope, then change administrator passwords and access keys.