CISA Adds Critical Ray Vulnerability to KEV Catalog Amid Active Exploitation
The U.S. Cybersecurity and Infrastructure Security Agency (CISA) has added a critical vulnerability affecting the Ray distributed computing framework to its Known Exploited Vulnerabilities (KEV) catalog after confirming evidence of active exploitation.
The vulnerability, tracked as CVE-2025-62593, has a CVSS score of 9.4 and can allow attackers to execute arbitrary code on vulnerable Ray environments through a browser based DNS rebinding attack.
What Is Ray?
Ray is an open source Python based distributed computing framework widely used to scale artificial intelligence and machine learning workloads.
The project has more than 43,500 GitHub stars and over 7,900 forks, making it a widely used platform in AI and machine learning environments.
How CVE-2025-62593 Works
The vulnerability is related to insufficient protection on critical Ray endpoints, including /api/jobs and /api/job_agent/jobs/.
Ray historically did not require authentication on these endpoints. Attackers can abuse this design together with a DNS rebinding attack to bypass browser security protections and interact with a vulnerable Ray instance.
The attack can be triggered when a developer running Ray visits a malicious website or views an advertisement controlled by an attacker.
The malicious webpage can abuse the victim's browser as an intermediary to send requests to the Ray service and ultimately execute arbitrary shell commands on the affected machine.
Developers Are a Key Target
The vulnerability primarily affects developers and organizations running Ray in development or testing environments.
A successful attack may allow a threat actor to execute arbitrary commands with the privileges of the Ray process. This could result in data theft, system compromise, malware deployment, or further movement within the victim's network.
The attack can also potentially target Ray instances located inside private corporate networks by using the victim's browser as a confused deputy.
Vulnerability Fixed in Ray 2.52.0
The Ray development team addressed the vulnerability in Ray version 2.52.0.
Security researcher Avi Lumelsky of Oligo was credited with discovering the fetch bypass, while Jonathan Leitschuh developed the DNS rebinding attack technique used to demonstrate exploitation.
Organizations running Ray should upgrade to version 2.52.0 or a later release as soon as possible.
Evidence of Active Exploitation
CISA has not publicly disclosed technical details about the attacks it observed in the wild.
However, previous research has linked CVE-2025-62593 to malicious activity.
A BitSight report published in March 2026 found that operators behind the RondoDox DDoS botnet had added the vulnerability to their attack arsenal shortly before its public disclosure. The availability of a proof of concept reportedly helped attackers incorporate the flaw into their operations.
Researchers at Oligo have also reported attacks against vulnerable Ray instances as part of a campaign known as ShadowRay 2.0.
The campaign targeted unpatched Ray clusters, including systems equipped with NVIDIA GPUs, with the goal of turning compromised infrastructure into a self replicating cryptocurrency mining botnet.
CISA Sets August 20 Deadline
Following the addition of CVE-2025-62593 to the KEV catalog, Federal Civilian Executive Branch agencies have been directed to apply the necessary security updates and mitigations by August 20, 2026.
Organizations using Ray should identify exposed instances, upgrade to Ray 2.52.0 or later, restrict access to Ray management endpoints, and avoid exposing development Ray environments directly to the internet.
