Skip to content
Threat Feed
high advisory

Unauthenticated Sensitive Information Exposure in WordPress AI Chatbot Plugin

An authentication bypass vulnerability in the AI Chatbot & Workflow Automation WordPress plugin allows unauthenticated attackers to exfiltrate API keys and task configurations via sequential ID enumeration.

The AI Chatbot & Workflow Automation plugin for WordPress (versions 1.4.6 and below) contains a critical security flaw resulting in sensitive information exposure. The vulnerability exists within the getCurrentTaskResults() method located in modules/workspace/controller.php, which fails to implement necessary authentication or authorization checks.

The plugin registers AJAX actions using the wp_ajax_nopriv_ hook, making them globally accessible to unauthenticated users. Furthermore, the method is excluded from the plugin's getNoncedMethods() array, and the base getPermissions() check returns an empty array. Attackers can exploit this by sending crafted AJAX requests to the vulnerable endpoint, iterating through task IDs to retrieve stored task objects. These objects contain plaintext OpenAI API keys, AI prompts, keywords, and specific model configuration parameters. This exposure allows threat actors to hijack third-party AI services linked to the compromised WordPress installation and potentially leverage the leaked configuration to refine further downstream attacks.

Attack Chain

  1. Attacker identifies a WordPress site running the AI Chatbot & Workflow Automation plugin.
  2. Attacker probes the site to determine if the wp-admin/admin-ajax.php endpoint is accessible for plugin-specific actions.
  3. Attacker crafts a series of HTTP GET or POST requests directed at the getCurrentTaskResults AJAX action.
  4. Attacker performs sequential enumeration of task IDs (e.g., iterating integer values) within the request parameters.
  5. The plugin processes the requests, bypassing authentication due to the use of wp_ajax_nopriv_.
  6. The plugin retrieves sensitive task data, including plaintext API keys, from the backend database.
  7. The sensitive data is returned to the attacker in a JSON-formatted response.
  8. Attacker parses the JSON response to harvest OpenAI API keys and configuration details for unauthorized usage.

Impact

Successful exploitation leads to the unauthorized disclosure of sensitive plugin configuration data, specifically plaintext OpenAI API keys. In an enterprise or high-traffic environment, this results in the theft of proprietary AI prompts and costly API credit theft. Unauthorized access to these keys could allow attackers to perform actions on behalf of the organization within the associated AI service platform.

Recommendation

  • Update the AI Chatbot & Workflow Automation plugin to a version beyond 1.4.6 immediately.
  • Audit access logs for high-frequency requests to admin-ajax.php containing the action parameter associated with getCurrentTaskResults.
  • If the plugin cannot be updated, disable the plugin until a patch is applied.
  • Revoke and rotate any OpenAI API keys stored in WordPress plugins that were exposed or suspected of being accessed by unauthorized parties.

Immediate actions

Deploy webserver detection rule to identify exploitation attempts.

Detection Engineering 24h

Threat Hunt

Search logs for any occurrence of 'getcurrenttaskresults' in the URI query string from unauthenticated sessions.

T1552.001 high high confidence hunt now

Data: Web server logs

Mitigations

Update plugin to version 1.4.7 or higher or disable the plugin.

immediate IT Operations

CVE-2026-6639

Detection coverage 1

Detect CVE-2026-6639 Exploitation - Unauthorized Access to getCurrentTaskResults

high

Detects exploitation attempts against the WordPress AI Chatbot & Workflow Automation plugin by monitoring for unauthenticated AJAX requests to the vulnerable task results endpoint.

sigma tactics: initial_access techniques: T1592 sources: webserver

Detection queries are available on the platform. Get full rules →