CVE DatabaseCVE-2025-62593
한국어 권고문 →
CVE-2025-625938.8/HIGHCISA KEVCWE-94CWE-352

Ray AI Compute Engine DNS Rebinding Remote Code Execution Vulnerability (CVE-2025-62593)

Affected Vendors
ray-project
Affected Components
ray
Published Date
2025-11-26

Executive & SecOps Summary

SECOPS ASSESSMENT
Affected Scope

Developers running Ray AI compute engine versions prior to 2.52.0 in local or development environments

Impact

Remote Code Execution (RCE) on the developer's machine via DNS rebinding and browser-based request forgery when visiting malicious websites or viewing malvertisements

Mitigation

Immediately upgrade the Ray package to version 2.52.0 or higher

Affected Products & Version Scope

All Ray versions prior to 2.52.0 (exploitable via Firefox and Safari when run as a local development tool)

Remediation & Live Workaround Commands

PATCH & WORKAROUND

1. Official Patch Application (Recommended)

Upgrade your Ray framework to the patched version (2.52.0 or higher).

bash
# If using pip
pip install --upgrade ray>=2.52.0

# Update your requirements.txt or pyproject.toml
ray>=2.52.0

2. Temporary Mitigations

If an immediate upgrade is not feasible:

  • Network Isolation: Restrict the Ray dashboard and API services to bind only to localhost (--host 127.0.0.1) and avoid exposing them to broader networks.
  • DNS Rebinding Protection: Use secure DNS servers or local router configurations that block external DNS responses containing private/loopback IP addresses (RFC 1918 filtering).
Recommended Security Control & Mitigation Partner
Datadog Cloud & Kubernetes Security
Due to active in-the-wild exploitation, immediately upgrade Ray to version 2.52.0 or higher to secure your development environments.

Detect Container Escape & Privilege Escalation Across Your Kubernetes Fleet

Continuous eBPF-based runtime threat detection and posture management for AWS, GCP, Azure, Docker, and Kubernetes workloads.

Real-Time Production Container CVE Mapping

Technical Root Cause & Exploitation Mechanics

A critical Remote Code Execution (RCE) vulnerability has been identified in the Ray AI compute engine. This vulnerability stems from an insufficient defense mechanism designed to guard against browser-based attacks.

Ray previously attempted to block unauthorized browser requests by validating if the User-Agent header started with the string "Mozilla". However, this defense is fundamentally flawed because the Fetch specification allows the User-Agent header to be manipulated or bypassed under certain conditions.

By combining this bypass with a DNS Rebinding attack, an attacker can exploit a developer running Ray locally. If the developer inadvertently visits a malicious website or is served a malicious advertisement (malvertising), the attacker's domain rebinds to the developer's localhost IP. This allows the browser to bypass the Same-Origin Policy (SOP) and send arbitrary API requests to the local Ray instance, leading to arbitrary code execution (CWE-94, CWE-352). This vulnerability is actively exploited in the wild and is listed in the CISA KEV catalog.

Official Vendor Advisories & References

NVD Original English Description

Ray is an AI compute engine. Prior to version 2.52.0, developers working with Ray as a development tool can be exploited via a critical RCE vulnerability exploitable via Firefox and Safari. This vulnerability is due to an insufficient guard against browser-based attacks, as the current defense uses the User-Agent header starting with the string "Mozilla" as a defense mechanism. This defense is insufficient as the fetch specification allows the User-Agent header to be modified. Combined with a DNS rebinding attack against the browser, and this vulnerability is exploitable against a developer running Ray who inadvertently visits a malicious website, or is served a malicious advertisement (malvertising). This issue has been patched in version 2.52.0.

SecOps Advisory Dispatch & RSS Feed

Subscribe to Daily Vulnerability Briefings & KEV Alerts

Concise morning briefings for SecOps, DevSecOps, and SRE teams covering newly exploited CVEs and mitigation priorities.

Slack / Teams / RSS Feed: