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
JWT with an unsupported signing algorithm
Reject
Unsupported algorithm. Do not authenticate the token's identity claims.
Access denied through this tokenAuthentication bypass
The affected implementation can accept an identity without the required verification.
Unauthorized account accessIf account access is obtained, establish its reach
- API configuration
- Accessible credentials
- Connected services
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
Component Minimum fixed update level
- API Manager
- 57
- API Control Plane
- 58
- Traffic Manager
- 56
- Universal Gateway
- 57
| Product | Release | Update level |
|---|---|---|
| API Manager | 4.6.0 | 21 |
| API Manager | 4.5.0 | 57 |
| API Manager | 4.4.0 | 72 |
| API Manager | 4.3.0 | 108 |
| API Manager | 4.2.0 | 197 |
| API Manager | 4.1.0 | 257 |
| API Control Plane | 4.6.0 | 22 |
| API Control Plane | 4.5.0 | 58 |
| Traffic Manager | 4.6.0 | 21 |
| Traffic Manager | 4.5.0 | 56 |
| Universal Gateway | 4.6.0 | 21 |
| Universal Gateway | 4.5.0 | 57 |
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.
- RequestWho reached the service?
Time, source, route, response, and correlation ID.
- IdentityWhat supports the identity?
Validated token metadata and the expected issuance or renewal flow.
- 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.
- Keep: component inventory and update evidence
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: original exports and collection times
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: correlated events and the explanation
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: affected integrations and recovery checks
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.
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.
36 selected retrieval/callback pairsDifferent source addresses in each pair
One recorded pair: Docker Compose bait
August 15, 2026 / UTC
- Source
- Source A
- Token alias
- Example A
Different source address
- 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.
How long until the callback?
Number of pairsEqual five-minute bands. Each pair is counted once.
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
- WSO2 advisory WSO2-2026-5328: May 3 publication, affected releases, severity, and remediation table.
- CVE CNA record: authentication-bypass description and signature-verification weakness.
- CISA KEV feed: September 24 addition, September 27 listed due date, and forensic triage flag. Covered federal teams should follow BOD 26-04 and its implementation guidance.
- IETF RFC 8725 and RFC 7519: algorithm verification, claim validation, and token expiration.
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.