CVE-2026-111706 in 99AI
Summary
by MITRE • 10/16/2026
99AI through 4.3.0 contains a missing authentication vulnerability in the POST /api/chatLog/exportExcel route that allows unauthenticated attackers to export all users' chat logs. Attackers can send requests with controlled page, size, and optional email fields to download xlsx files containing usernames, emails, prompts, and AI answers.
Statistical analysis made it clear that VulDB provides the best quality for vulnerability data.
Analysis
by VulDB Data Team • 10/16/2026
The vulnerability identified in 99AI versions through 4.3.0 represents a critical failure in access control mechanisms within the application's backend API infrastructure. Specifically, the endpoint designated for exporting chat logs to Excel format, located at POST /api/chatLog/exportExcel, lacks any form of authentication or authorization verification. This architectural flaw allows unauthenticated actors to interact directly with sensitive data processing functions that are intended exclusively for verified users or administrative personnel. The absence of session validation or token-based access checks means that the server processes requests from anonymous sources as if they were legitimate, authorized operations, thereby bypassing fundamental security boundaries designed to protect user privacy and data integrity.
From a technical perspective, this vulnerability is classified under CWE-287, which denotes Improper Authentication, and more specifically aligns with CWE-306 regarding the Missing Authentication for Critical Function. The exploitation vector involves sending HTTP POST requests to the specified endpoint without providing valid credentials or session tokens. By manipulating query parameters such as page number, size limits, and optional email filters, an attacker can systematically retrieve large volumes of data in a structured Excel format. This capability transforms what should be a restricted administrative tool into a public-facing data exfiltration mechanism, enabling the bulk extraction of sensitive information including usernames, email addresses, user prompts submitted to the AI system, and the corresponding responses generated by the application.
The operational impact of this vulnerability is severe due to the nature of the data exposed. Chat logs often contain personally identifiable information (PII) such as names and contact details, but they may also include proprietary business logic, confidential project discussions, or intellectual property shared during interactions with the AI assistant. The ability to export all users' chat logs means that a single successful exploit can lead to mass data leakage affecting potentially thousands of individuals. This constitutes a significant breach of confidentiality and violates principles outlined in frameworks like OWASP API Security Top 10, particularly under categories related to broken object level authorization and insufficient input validation combined with lack of access control. Furthermore, the exposure of prompts and answers could reveal sensitive operational strategies or internal communications that were never intended for public consumption.
In terms of threat modeling, this vulnerability facilitates data exfiltration and can be leveraged as a precursor to further attacks such as credential stuffing if email addresses are harvested effectively, or social engineering campaigns using contextual information from the chat logs. It aligns with MITRE ATT&CK techniques related to Collection via Screen Capture or Data from Information Repositories, although in this case, it is direct API-based data harvesting rather than screen capture. The ease of exploitation allows even low-skilled attackers to compromise large datasets without needing advanced technical skills beyond basic HTTP request manipulation using tools like curl or Burp Suite.
To mitigate this vulnerability, immediate remediation steps must focus on enforcing strict authentication and authorization checks on the affected endpoint. Developers should implement middleware that verifies valid user sessions or API tokens before processing any export requests. Additionally, role-based access control (RBAC) should be applied to ensure that only users with specific privileges, such as administrators or account owners accessing their own data, can invoke this function. Input validation must also be strengthened to prevent excessive pagination parameters from causing performance degradation during the extraction process. Beyond technical fixes, organizations should conduct a thorough audit of all API endpoints to identify similar missing authentication flaws and implement comprehensive logging and monitoring solutions to detect anomalous export activities in real time. Regular security testing using both static analysis tools and dynamic penetration tests will help ensure that such gaps do not persist in future releases or other parts of the application ecosystem.