Summary
The Payload Alt-Text Plugin allows any authenticated user accepted by its default guard to read protected upload documents and overwrite their alt and keywords fields. This becomes risky when the target collection relies on administrator-only read or update access rules, because the plugin's Local API calls use Payload's default access-override behavior.
Affected context
The issue occurs in deployments using the Payload Alt-Text Plugin endpoints for single or bulk alt-text generation. It requires a target collection with meaningful read or update access controls and an authenticated account that can reach the plugin endpoints. The risk is present when the endpoint invokes Payload document operations without explicitly disabling access overrides.
Exploit preconditions
- Attack surface: The plugin's authenticated HTTP endpoints for single and bulk alt-text generation
- Required access: Any authenticated user permitted through the plugin's default guard
- Trigger point: A request that supplies a target document identifier, collection, locale, and update values
- Why exploitation is possible: The plugin calls document lookup and update operations without
overrideAccess: false, allowing Payload to bypass collection-level authorization functions
Why this happens
The underlying bug is an authorization-control mismatch between the plugin endpoint and the Payload Local API. Payload defaults overrideAccess to true when the option is not provided, so the plugin's document reads and updates can bypass the target collection's access functions. This maps to CWE-863, Improper Authorization, because a low-privilege authenticated principal can perform operations reserved for more privileged users.
Potential impact
- An attacker may read protected upload documents
- An attacker could overwrite alt-text and keyword metadata
- Unauthorized metadata changes can affect search, accessibility, or content workflows
- Modified metadata may cause content integrity or moderation issues
- Bulk operations could increase the scope and speed of unauthorized changes
General mitigation guidance
Mitigation generally focuses on preserving collection-level authorization
decisions across every plugin and Local API operation.
Common defensive patterns include:
- Explicitly disabling access overrides for operations handling user-controlled targets
- Applying authorization checks before reading or updating each document
- Restricting plugin endpoints to roles that legitimately require the operation
- Validating collection, document, locale, and update fields against an allowlist
- Limiting bulk-operation scope and requiring appropriate privilege for each target
- Using least-privilege service identities and separating content-generation permissions from administrative permissions
- Treating authenticated input as untrusted unless authorization has been independently verified
How teams detect issues like this
Organizations can look for this class of issue through:
- Static analysis rules identifying Payload
findByIDorupdatecalls that omit explicit access-control options - Code review checks for plugin endpoints that accept document identifiers, collections, or update fields from clients
- Tests that exercise endpoints with low-privilege accounts against administrator-protected collections
- Authorization regression tests verifying that denied Local API operations remain denied through plugin routes
- Runtime monitoring for unusual bulk metadata updates or access to collections outside a user's normal role
- Policy enforcement requiring explicit authorization decisions for privileged document operations
Automated security review tools may help identify missing authorization parameters, but application-level tests remain important for confirming effective access behavior.

