Oracle Access Manager CVE-2026-70905

Exploitation requires that the target Oracle Access Manager instance (12.2.1.4.0 or 14.1.2.1.0) exposes its SAML-processing Agent infrastructure endpoint to the attacker over the network - the description explicitly names 'network access via SAML' as the vector, so a deployment with SAML federation/agent handling reachable by the attacker is the concrete prerequisite. Per the CVSS vector (AV:N/AC:L/PR:N/UI:N), no authentication and no user interaction are needed against affected versions. The primary limiting factor is network reachability of the SAML endpoint: instances whose OAM authentication/federation surface is confined to internal or trusted segments substantially reduce exposure, and deployments not using the affected SAML/agent path present a narrower target, though defenders should confirm whether the vulnerable code path is active regardless of federation configuration.

This is a genuine top-priority issue, not a high-CVSS-but-low-real-risk case. The CVSS vector (AV:N/AC:L/PR:N/UI:N) describes trivially reachable, unauthenticated, no-interaction network exploitation with full C/I/A impact, and Oracle explicitly labels it 'easily exploitable' resulting in 'takeover' - the qualitative language reinforces the numeric 9.8. No EPSS score, KEV entry, SSVC decision, or POC is present in the supplied data, so likelihood-of-exploitation telemetry is unavailable; absence of KEV means no confirmed active exploitation, but that should not lower urgency given the target is an identity/SSO chokepoint whose compromise defeats authentication for all fronted applications. One notable tension to flag: Oracle's tag 'Information Disclosure' conflicts with the vector's I:H/A:H and the 'takeover' outcome - treat the CVSS/description ('takeover') as authoritative over the narrower tag. CWE is N/A, so root-cause classification is inferred, not confirmed.

An attacker on the network sends a crafted SAML message to an internet-reachable Oracle Access Manager Agent/federation endpoint, exploiting the assertion-handling flaw to authenticate as or impersonate a trusted principal without any credentials. With OAM compromised, the attacker mints valid sessions for every application the SSO layer protects, achieving broad lateral access. No public exploit identified at time of analysis, and the attack requires no user interaction.