CISA Flags Actively Exploited Ray Flaw That Can Trigger Browser-Based RCE

WASHINGTON, 18 AUGUST 2026 —

CRITICAL RAY FLAW ADDED TO U.S. EXPLOIT CATALOG AS ACTIVE ATTACKS DETECTED

Ray, a Python-native platform used to scale artificial intelligence and machine learning workloads, has gained significant traction with over 43,500 GitHub stars and 7,900 forks. The flaw, tracked as CVE-2025-62593 with a CVSS score of 9.4, enables remote code execution (RCE) via web browsers such as Mozilla Firefox and Apple Safari through a DNS rebinding attack.

The vulnerability stems from Ray’s longstanding decision not to implement authentication on critical endpoints like `/api/jobs` and `/api/job_agent/jobs/`. This oversight allows attackers to execute arbitrary code on systems running Ray, particularly in development environments. According to a November 2025 advisory by Ray maintainers, the issue is exacerbated when developers inadvertently visit malicious websites or are served malicious advertisements, enabling shell code execution on their machines.

The attack can also propagate to network-adjacent Ray instances by using the browser as a confused deputy, targeting Ray deployments within private corporate networks. Ray has since patched the flaw in version 2.52.0 of the Python package. The discovery was credited to Oligo security researcher Avi Lumelsky for the fetch bypass and Jonathan Leitschuh for the DNS rebinding technique.

CISA has not disclosed details of real-world exploitation methods. However, a March 2026 report by BitSight revealed that the threat actors behind the RondoDox DDoS botnet had integrated the vulnerability into their toolkit just two days before its public disclosure on November 26, 2025, following the release of a proof-of-concept (PoC) exploit.

Unpatched Ray instances have also been targeted in a campaign dubbed ShadowRay 2.0, where cybercriminals compromise clusters equipped with NVIDIA GPUs to build a self-replicating cryptocurrency mining botnet.

In response, the U.S. Federal Civilian Executive Branch (FCEB) agencies have been directed to apply necessary fixes and mitigations by August 20, 2026.

Impact on Malaysia’s Tech Sector

For Malaysia, where AI and cloud computing adoption is growing, the exploitation of Ray’s vulnerability poses significant risks, particularly for enterprises and research institutions relying on open-source AI frameworks. The country’s push toward digital transformation and AI integration in sectors such as finance, healthcare, and manufacturing increases exposure to such threats.

Local cybersecurity firms and cloud service providers may need to audit their Ray deployments and ensure immediate patching to prevent potential breaches. The incident underscores the broader challenge of securing open-source tools widely used in AI development, which often lack default authentication safeguards.

Sector and Company Implications

Ray’s widespread use in AI/ML workloads means the vulnerability affects a broad range of organizations, from startups to large corporations. The patch in version 2.52.0 is critical, but organizations running older versions remain at risk. The integration of the exploit into botnet campaigns like ShadowRay 2.0 highlights the financial motivations driving such attacks, particularly in cryptocurrency mining.

For Malaysian companies leveraging Ray, immediate action is advised: update to version 2.52.0 or later, enforce network segmentation, and monitor for unusual GPU activity indicative of cryptomining malware.

Outlook and Recommendations

The active exploitation of CVE-2025-62593 signals a broader trend where threat actors rapidly weaponize newly disclosed vulnerabilities, often within days. This underscores the need for proactive patch management, threat intelligence sharing, and authentication controls in open-source AI tools.

For Malaysian businesses, collaboration with local cybersecurity agencies such as the National Cyber Security Agency (NACSA) and adherence to CISA’s mitigation timelines will be in mitigating risks. As AI adoption accelerates, securing the underlying infrastructure—including distributed computing frameworks like Ray—will remain a critical priority to prevent operational disruptions and financial losses.

Related: Apple