Description

A vulnerability was identified in Doccano Open Source Annotation Tools for Machine Learning Practitioners and Auto Labeling Pipeline Module to Annotate a Document Automatically up to 1.8.5. Affected by this issue is the function ExampleDetail of the file /v1/projects/1/examples/ of the component Project Example Detail Endpoint. Such manipulation leads to improper access controls. The attack may be launched remotely. The exploit is publicly available and might be used. The vendor was contacted early about this disclosure but did not respond in any way.

Executive Summary

Technical Details

Mitigations

References

Additional Metadata

{
  "score": "6.3",
  "pubDate": "2026-08-31T20:17:15.337Z",
  "pubdate": "2026-08-31T20:17:15.337Z",
  "executiveSummary": "A critical improper access control vulnerability exists in Doccano versions up to 1.8.5 within the Project Example Detail Endpoint. This security flaw enables unauthorized remote actors to bypass existing access restrictions and interact with the Project Example Detail functionality without proper authorization.\nThe vulnerability resides within the ExampleDetail function located in the /v1/projects/1/examples/ file path. Because this endpoint fails to adequately validate the user's permissions or authorization context, an attacker can manipulate requests to gain unauthorized access to sensitive project data.\nThe risk implication is significant as it facilitates unauthorized information disclosure and potential data manipulation within the Doccano machine learning annotation environment. The vulnerability is exploitable remotely, and since exploit code is publicly available, the risk of exploitation is elevated. The vendor has not provided a response or a patch, leaving users exposed to active threats.\nAttackers do not require complex specialized access beyond network connectivity to the target instance to initiate the exploitation process, making this a high-priority concern for organizations managing private or sensitive datasets using this software.",
  "technicalDetails": "The vulnerability is an Improper Access Control flaw within the Doccano application's API layer, specifically affecting the Project Example Detail Endpoint. The root cause lies in the insufficient enforcement of identity-based authorization checks within the ExampleDetail function. In standard secure design, endpoints governing project-specific resources should verify that the requester possesses explicit membership or administrative permissions for the project ID requested in the URI path.\nIn the affected versions (up to 1.8.5), the /v1/projects/1/examples/ endpoint fails to validate whether the authenticated or unauthenticated session context is authorized to view or modify data associated with the project identifier. This indicates a failure in the application’s middleware or view-level authorization logic to properly inspect the relationship between the requesting user and the target project resource.\nThe attack flow commences with the remote actor identifying an active Doccano instance. By crafting a specific HTTP request directed at the /v1/projects/{id}/examples/ path, an attacker can bypass the intended authorization constraints. Since the ExampleDetail function lacks the necessary access control list (ACL) verification or object-level permission checks, it returns object details that should be restricted based on user privilege levels.\nExploitation is conducted remotely over standard HTTP/HTTPS protocols. Because the logic error exists in the server-side code, the server processes the request and leaks information back to the attacker in the response body. Given that the exploit is publicly available, an attacker can automate this reconnaissance and data retrieval process, leading to a large-scale exfiltration of training datasets or annotation records stored within the Doccano instance.\nThe impact post-exploitation includes unauthorized exposure of proprietary machine learning training data and potential metadata associated with annotation projects. There is no indication that the vulnerability requires a specific privilege level, suggesting that even low-privileged users or unauthenticated entities may be able to reach the vulnerable code path depending on the global API configuration. The absence of a vendor-provided patch necessitates immediate manual intervention to mitigate the risk of ongoing or future exploitation."
}