Known exploited API security

WSO2 JWT bypass: check what the account could reach

An authentication bypass can turn a forged identity into an apparently legitimate request. For an API platform, the investigation has to follow that access into application settings, credentials, and connected services.

Vendor severity
Critical CVSS 10.0, or 9.8 single-tenant
Prior privileges
None required PR:N
KEV listed due date
September 27 Covered federal systems

The federal requirements come from BOD 26-04 and its implementation guidance, including asset exposure and forensic triage. The catalog date is not a universal private-sector deadline. CISA directive

01 / The failed trust check

An unsupported signing algorithm must not become permission to enter

In affected WSO2 products, a JWT signed using an unsupported algorithm can bypass authentication. WSO2 identifies potential administrative account compromise and account takeover. The flaw affects API Manager and specified releases of API Control Plane, Traffic Manager, and Universal Gateway. WSO2 advisory

A JWT's identity claims are assertions until validation succeeds. The receiving application must enforce its permitted algorithms and verify the cryptographic operation. It must also apply the relevant issuer, audience, and lifetime checks. A readable token is not a verified identity. JWT security best practices

CVE-2026-5430 / Conceptual authentication pathThe trust decision happens before account access.
One incoming request

JWT with an unsupported signing algorithm

HeaderUnsupported algorithm
ClaimsAsserted identity
SignatureNot verified
Expected validation

Reject

Unsupported algorithm. Do not authenticate the token's identity claims.

Access denied through this token
Affected validation

Authentication bypass

The affected implementation can accept an identity without the required verification.

Unauthorized account access

Possible administrative account takeover, according to WSO2. Downstream reach depends on the account's permissions and the deployment.

For an organization using WSO2 to manage access to critical services, that account's reach is the operational concern. Identify the API configurations it can change and the credentials it can retrieve. A platform account compromise can create a path to other services, but it does not automatically establish access to every backend or to an industrial control system.

A discrepancy in the KEV description

As checked on September 24, CISA's entry labels this CVE as path traversal and describes file upload leading to code execution. Its linked WSO2 advisory and CNA record describe JWT authentication bypass, and the KEV weakness field lists CWE-347, improper verification of a cryptographic signature. Use the vendor's technical description for remediation and hunting. Keep the KEV scheduling and triage requirements in view. Current CISA feed · CNA record

02 / Verify the installed fix

The product version alone is not enough

The same release number can appear in both affected and updated installations. Subscription customers should verify the update level for each deployed component against the table below. Community users should apply the public fixes linked by WSO2 or move to a vendor-confirmed unaffected release. Vendor remediation instructions

Why a version-only check falls shortOne release number. Component-specific fixes.
Release4.5.0

Component Minimum fixed update level

API Manager
57
API Control Plane
58
Traffic Manager
56
Universal Gateway
57
WSO2's minimum fixed subscription update levels. Higher levels also qualify.
ProductReleaseUpdate level
API Manager4.6.021
API Manager4.5.057
API Manager4.4.072
API Manager4.3.0108
API Manager4.2.0197
API Manager4.1.0257
API Control Plane4.6.022
API Control Plane4.5.058
Traffic Manager4.6.021
Traffic Manager4.5.056
Universal Gateway4.6.021
Universal Gateway4.5.057

Record the product, release, update level or applied community fix, and verification time for every relevant node. An updated gateway does not establish that a separately deployed control-plane component has also been fixed.

03 / Account for the access

A successful request deserves more attention than a failed login count

I would have the API platform owner and the response team work from the same inventory. The platform owner knows which accounts and integrations are expected. The response team needs that context to distinguish routine automation from privileged activity that cannot be explained.

Correlate the request with its identity and subsequent actions
  1. RequestWho reached the service?

    Time, source, route, response, and correlation ID.

  2. IdentityWhat supports the identity?

    Validated token metadata and the expected issuance or renewal flow.

  3. ActivityWhat did the account do?

    Privileged reads, configuration changes, and related service access.

Prioritize a suspicious authentication result followed by sensitive activity. These are investigation leads, not a published campaign signature.

  1. Limit exposure while confirming the update

    Restrict management access to approved administration paths. Review reachability of affected API services as well. Network restrictions reduce exposure but do not repair token validation. Install the appropriate vendor fix and verify each node's running state.

    Keep: component inventory and update evidence
  2. Preserve the records before they disappear

    Retain available application audit logs, identity-provider records, and reverse-proxy or gateway access logs. Preserve relevant configuration and clock information before disruptive recovery where practical. Base the review window on actual exposure and retention, not just the KEV addition date.

    Keep: original exports and collection times
  3. Review privileged API activity against expected use

    Investigate unexpected administrative actions and token-validation anomalies. Where safe metadata is already available, compare the declared algorithm, issuer, and audience with the application's expected policy. An administrator request without a nearby interactive login is a lead, not proof: service accounts and token renewal can legitimately produce that pattern.

    Keep: correlated events and the explanation
  4. Scope credential recovery to the access at risk

    Identify application secrets and backend credentials the affected account could retrieve. If compromise is confirmed or exposure cannot be bounded with the available evidence, agree a rotation and revocation plan with the application owners. Secure the platform first so replacement secrets are not exposed through the same access path.

    Keep: affected integrations and recovery checks

Do not create a new secret store in the logs. Capture authentication decisions and appropriately minimized metadata. Avoid writing full bearer tokens, authorization headers, or application secrets into general-purpose monitoring.

04 / Detection beyond the authentication result

Instrument the resources a compromised identity would seek

An apparently authenticated request can still be hostile. Alongside the platform's audit trail, a deliberately placed decoy resource or decoy credential can provide another investigation lead when someone interacts with it. Define what should trigger, how the alert reaches the existing monitoring workflow, and who follows it up. Validate that path with a controlled test.

Separate first-party observation / August 2 to 29, 2026A retrieval became a second event to investigate.

36 selected retrieval/callback pairsDifferent source addresses in each pair

One recorded pair: Docker Compose bait

August 15, 2026 / UTC

01 / Bait retrieved
Source
Source A
Token alias
Example A
02 / HTTP callback
Source
Source B
Token alias
Example A

The token association connects the two events. The elapsed time and changed source give the response team a concrete lead to examine.

Highlight timing is compressed. Timestamps are recorded.

How long until the callback?

Number of pairs
  1. 0 to <5 min12
  2. 5 to <10 min2
  3. 10 to <15 min17
  4. 15 to <20 min3
  5. 20 to <25 min2

Equal five-minute bands. Each pair is counted once.

Median across the 36 pairs12.6 min12m 37s exactly

Observed range
30s to 24m 31s

These are HTTP callbacks associated with issued bait, not credential authentications or WSO2 exploit captures. Times were checked against the raw records. Source and token labels are aliases. Earlier analysis and cohort details

For this deployment, keep decoy material separate from real backend credentials and give it no production access. Test its placement with the API owner so normal discovery and administrative jobs do not become an unexplained stream of alerts.

The patch record answers whether the vulnerable code is fixed. The investigation should answer what an affected identity could reach and whether that access needs recovery.

Sources and review scope

Reviewed September 24, 2026. The honeytoken figure uses the August cohort first reviewed September 5. All 36 pairs and their elapsed times were independently matched to raw records on September 24. The selected cohort is not a callback rate or a fleet-wide sample. The WSO2 findings come from the vendor and CISA.