Summary
Authorizer versions before 2.2.1 may accept attacker-controlled redirect URIs at the /authorize endpoint without checking them against configured allowed origins. For token-based response types, this can cause access tokens, identity tokens, and refresh tokens to be sent to an attacker-controlled destination, potentially enabling account or session compromise.
Affected context
The vulnerability occurs in Authorizer deployments where the /authorize endpoint processes OAuth or OpenID Connect requests and does not validate the supplied redirect_uri against an approved origin list. It is especially risky when the endpoint supports response types that place tokens in redirect parameters and when an attacker can construct authorization requests using a valid public client identifier.
Exploit preconditions
- Attack surface: Public
/authorizeendpoint - Required access: No authentication is generally required; a valid client identifier can be obtained from publicly exposed application metadata
- Trigger point: Processing an authorization request with an attacker-controlled
redirect_uriand a token-bearing response type - Why exploitation is possible: The endpoint may redirect to an untrusted destination while including authorization artifacts in the redirect parameters
Why this happens
The underlying bug is insufficient validation of a security-sensitive redirect target. The /authorize handler accepts a caller-supplied redirect_uri without enforcing an allowlist or equivalent exact-match policy tied to the registered client. When token-bearing response types are used, redirecting to that untrusted location can disclose sensitive authorization material. This behavior maps to CWE-601: URL Redirection to Untrusted Site ('Open Redirect').
Potential impact
- An attacker could receive access tokens, identity tokens, or refresh tokens
- User accounts or sessions may be compromised
- Authorization responses could be redirected to phishing or credential-harvesting infrastructure
- Confidentiality and integrity of protected resources may be affected
- Applications relying on the authorization server could experience broader trust-boundary compromise
General mitigation guidance
Mitigation generally focuses on making redirect validation strict, centralized, and secure by default.
Common defensive patterns include:
- Permit only redirect URIs registered for the specific client
- Prefer exact matching over loose prefix, substring, wildcard, or pattern matching
- Validate scheme, host, port, path, and encoding consistently before issuing any redirect
- Reject unrecognized or malformed redirect targets rather than falling back to user-supplied values
- Avoid placing bearer tokens in URLs where safer authorization-code flows are available
- Minimize token scope, lifetime, and privileges, and use sender-constrained or otherwise protected tokens where feasible
- Apply the same validation policy across every authorization and account-action handler
- Protect client metadata and prevent unauthorized changes to registered redirect URIs
How teams detect issues like this
Organizations can detect this vulnerability class through:
- Static analysis rules that identify redirects based directly on request parameters
- Code review checks for missing allowlist validation around
redirect_uri - Tests covering unregistered hosts, subdomains, alternate ports, encoded URLs, and unexpected schemes
- Dynamic testing that verifies token-bearing responses never reach unapproved origins
- Policy enforcement requiring exact redirect-URI registration for every OAuth or OpenID Connect client
- Monitoring for authorization requests and redirects to domains outside the approved application inventory
Automated security review tools may help identify untrusted redirect flows, but endpoint-specific tests are useful for confirming token disclosure risks.

