The U.S. Cybersecurity and Infrastructure Security Agency (CISA) has added a critical vulnerability to its Known Exploited Vulnerabilities (KEV) catalog.
This vulnerability, tracked as CVE-2025-62593, is a code injection flaw in the Ray Project, a widely used open-source distributed computing framework often deployed for artificial intelligence workloads, machine learning development, data processing, and scalable Python applications. CISA has warned that this issue is currently being actively exploited.
Ray-Project Ray Code Injection Vulnerability
Successful exploitation of this vulnerability could allow remote code execution, enabling threat actors to run commands under their control on vulnerable systems.
CVE-2025-62593 is classified as a code injection vulnerability under CWE-94 and is also associated with CWE-352, which covers cross-site request forgery.
The flaw may particularly affect developers who use Ray as a development tool, especially in environments involving Firefox and Safari. While CISA has not linked this vulnerability to any ransomware activities, the active exploitation significantly heightens the urgency for organizations using Ray.
CISA added this vulnerability to its KEV catalog on August 17, 2026, and set a remediation deadline of August 20, 2026.
Under Binding Operational Directive 26-04, federal civilian executive branch agencies are required to apply mitigations by that date, prioritizing security updates based on real-world exploitation risks. The agency also urged organizations to follow relevant forensic triage guidelines when assessing potentially affected systems.
Ray deployments can expose organizations to significant risk, especially when dashboards, APIs, development environments, or distributed compute clusters are accessible from untrusted networks.
A successful attack could grant attackers access to sensitive workloads, credentials, source code, cloud resources, or connected internal infrastructure.
In AI and data science environments, this vulnerability could also facilitate model theft, data exfiltration, workload manipulation, and lateral movement within networks.
Organizations should promptly check if Ray is installed in development workstations, CI/CD systems, cloud instances, Kubernetes clusters, research infrastructure, or externally accessible computing environments.
Security teams need to identify exposed Ray components, implement vendor-provided mitigations, restrict access to management interfaces, and eliminate unnecessary internet exposure.
Administrators should inspect logs for unusual request activity, unexpected browser interactions, abnormal process execution, newly created user accounts, and suspicious outbound network connections.
Given that this flaw could allow remote code execution, incident responders should treat any signs of exploitation as indicators of a full system compromise.
CISA recommends that organizations follow vendor guidance and comply with applicable BOD 26-04 requirements for cloud services. If no mitigation is available, organizations should discontinue the use of affected products.
Stop new phishing & malware before they compromise your business. Integrate live intel from 15K SOCs around the world
 has added a critical vulnerability to its Known Exploited Vulnerabilities (KEV) catalog.){kind=link}