/

CVE-2025-61164

Severity:

HIGH RISK

(7.5)

Information Disclosure (CWE-200) in Cohere North WebSocket

CVE-2025-61164

Deferred

Vulnerability type

Category:

Information Disclosure

CWE:

CWE-200

Impact:

Unauthorized Information Exposure

Severity:

HIGH RISK

(7.5)

CVSS version:

3.1

Catch this CVE while code is being written

Detect vulnerable code directly in pull requests, so CVEs are fixed before merge.

Summary

Cohere North AI v1.1.5 contains an information disclosure vulnerability in its WebSocket endpoint. Under conditions where the endpoint exposes data without adequately enforcing access controls or response isolation, an attacker may obtain information that should remain restricted.

Affected context

The issue occurs in deployments of Cohere North AI v1.1.5 that expose the affected WebSocket endpoint. Risk depends on how the endpoint handles authentication, authorization, connection state, and the data returned over WebSocket messages.

Exploit preconditions

  • Attack surface: The WebSocket endpoint
  • Required access: Ability to reach or establish a connection to the endpoint; additional authentication requirements may depend on deployment configuration
  • Trigger point: Processing a WebSocket connection or request that causes information to be returned
  • Why exploitation is possible: Inadequate access control or response isolation may allow data to be disclosed to an unauthorized client

Why this happens

The underlying bug is an information exposure condition in WebSocket request handling. Data may be returned without sufficiently verifying that the connected client is authorized to receive it or that the response is limited to that client's permitted scope. This maps to CWE-200: Exposure of Sensitive Information to an Unauthorized Actor.

Potential impact

  • Attackers may access sensitive application or session information
  • Exposed data could support further attacks against users or services
  • Confidentiality of messages or WebSocket-related state may be reduced
  • Organizations could experience privacy, compliance, or operational consequences

General mitigation guidance

Mitigation generally focuses on reducing trust in client-controlled WebSocket state and enforcing authorization at the point where data is selected and returned.

Common defensive patterns include:

  • Applying authentication and per-resource authorization to WebSocket connections and messages
  • Returning only data explicitly permitted for the authenticated principal
  • Isolating connection state between users, tenants, and sessions
  • Avoiding sensitive data in responses unless it is required for the operation
  • Using least-privilege service identities and narrowly scoped data access
  • Enforcing secure defaults for unauthenticated, expired, or malformed connections

How teams detect issues like this

Organizations can look for this class of issue through:

  • Static analysis and code review checks for WebSocket handlers that return data before authorization
  • Review of connection lifecycle, tenant isolation, and session-to-user mapping
  • Tests that attempt to access another user's or tenant's data through an otherwise valid connection
  • Runtime monitoring for unusual WebSocket access patterns, repeated enumeration, or responses exceeding expected scope
  • Policy enforcement requiring authorization decisions at every sensitive data access point

Sources

SourceLink
cohere.comcohere.com/north
github.comgithub.com/bdadoa/Cohere---North-AI-1.1.5---Vulnerabilities/blob/main/CVE-2025-61164.md

Frequently asked

questions

It may be remotely exploitable when the affected WebSocket endpoint is reachable by an attacker and returns information without adequate authorization.

Not necessarily. The required access depends on the endpoint's deployment and authentication controls; a valid account may still be insufficient if authorization boundaries are weak.

Enforce authentication and fine-grained authorization before selecting or returning WebSocket data, while isolating connection and tenant state.

Include negative authorization tests in application and integration testing, minimize sensitive responses, and monitor WebSocket traffic for unauthorized access patterns.

Find out what's already
exploitable in your codebase.