Loading story…
CISA added CVE-2025-62593 to its Known Exploited Vulnerabilities catalog after confirming active exploitation of a critical flaw in the open-source distributed computing framework Ray. The vulnerability is a code injection issue that can lead to remote code execution through a DNS rebinding attack involving browsers including Mozilla Firefox and Apple Safari, with developers running Ray in development or testing environments facing the greatest exposure. CISA updated the KEV entry to reflect an earlier dateAdded and moved the federal remediation deadline up to August 20, 2026.
Ray maintainers said the issue was caused by unauthenticated critical endpoints and inadequate protections against browser-based attacks, and they released a fix in Ray 2.52.0. Reporting tied the flaw to multiple real-world abuse cases before and after disclosure, including weaponization by the RondoDox DDoS botnet operators and separate exploitation of unpatched Ray instances in ShadowRay 2.0, a cryptocurrency-mining campaign observed by Oligo. CISA said known ransomware use is unknown and advised organizations to apply vendor mitigations and follow its remediation and forensic triage guidance.
Mallory correlates global threat intelligence with your attack surface — know if you’re exposed before adversaries strike.
6 events from the most recent confirmed update back to the earliest known activity.
CISA updated its Known Exploited Vulnerabilities catalog on 2026-08-17 to add or update CVE-2025-62593, identifying it as a Ray code injection flaw with evidence of active exploitation. The catalog entry notes exposure through Firefox and Safari and directs organizations to apply vendor mitigations and follow BOD 26-04 guidance.
In March 2026, BitSight reported that RondoDox operators had incorporated CVE-2025-62593 into their botnet toolkit before the vulnerability's public disclosure. The report highlighted how quickly the flaw was operationalized in the wild.
A BitSight report cited in the coverage said RondoDox DDoS botnet operators added CVE-2025-62593 to their arsenal two days before the flaw was publicly disclosed on November 26, 2025. The report attributed the rapid weaponization in part to the availability of a proof-of-concept exploit.
Ray maintainers disclosed the Ray code injection vulnerability in a security advisory shared in November 2025. The advisory described browser-based attack conditions tied to unauthenticated critical endpoints and insufficient protections against DNS rebinding-style abuse.
Oligo observed separate attacks abusing unpatched Ray deployments to turn NVIDIA GPU-equipped clusters into a self-replicating cryptocurrency-mining botnet dubbed ShadowRay 2.0. The article does not anchor a specific date for when those attacks were observed.
Ray maintainers released a fix for CVE-2025-62593 in Ray version 2.52.0. The KEV entry also references a Ray patch commit associated with the remediation.
Vulnerabilities, threat actors, malware, products, organizations, and breaches Mallory has linked to this story.
6 references tracked. Mallory keeps watching after this page renders.
theregister.com
Open sourcecybersecuritynews.com
Open sourcesecurityaffairs.com
Open sourcethehackernews.com
Open sourcegithub.com
Open sourcecve.org
Open sourceMap indicators from this story to your assets and identify affected systems in minutes.
Every observed campaign, victim, and pivot linked to actors named in this story.
Malware, exploits, and IOCs connected to the activity described here.
YARA, Sigma, and Snort rules deployed to your SIEM as soon as they’re published.
Get matching new stories delivered to your team as they break — not the next morning.
Ask questions about this story and take action on the answers.
Other stories Mallory is tracking that overlap on entities, sources, or topic labels.
A critical remote code execution (RCE) vulnerability, tracked as CVE-2025-62593, was discovered in the Ray AI compute engine, affecting developers using Ray as a development tool. The flaw is exploitable through Safari and Firefox browsers due to insufficient browser-based attack defenses, specifically relying on the `User-Agent` header for protection, which can be bypassed according to the fetch specification. Attackers can leverage DNS rebinding techniques to exploit this vulnerability if a developer visits a malicious website or is exposed to malvertising, potentially allowing remote code execution on the developer's system. The vulnerability has been addressed in Ray version 2.52.0, and users are urged to update to this version to mitigate the risk. Security researchers highlighted that the vulnerability is particularly dangerous because it targets developers in their local environments, making it possible for attackers to gain unauthorized access and control. The issue was rated as critical with a CVSS score of 9.4, underscoring the severity of the threat. The Ray project has released security advisories and patches, and organizations using Ray in development should prioritize applying the update to prevent exploitation through browser-based attacks on Safari and Firefox.
Jun 29, 2026
A global cybercriminal campaign, dubbed ShadowRay 2.0, is actively exploiting a critical unpatched vulnerability (CVE-2023-48022) in the open-source Ray distributed computing framework, which is widely used for AI workloads. Attackers, tracked as IronErn440, are leveraging the Ray dashboard API to achieve unauthenticated remote code execution, allowing them to hijack exposed Ray clusters and convert them into a self-replicating botnet. This botnet is used for cryptocurrency mining, data and credential theft, and launching distributed denial-of-service (DDoS) attacks. The campaign represents a significant evolution from previous ShadowRay attacks, with new techniques and AI-generated payloads observed, and is believed to have been active since at least September 2024. Despite the severity of the vulnerability, which carries a CVSS score of 9.8, no patch has been issued, as the original vendor, Anyscale, maintains Ray is intended for use in strictly controlled environments. The attack surface has grown dramatically, with over 200,000 Ray servers now exposed to the internet, many belonging to startups, research labs, and cloud-hosted AI environments. The attackers have demonstrated advanced evasion tactics, such as limiting CPU usage and disguising malicious processes, and have migrated their payload delivery infrastructure from GitLab to GitHub. The Ray project has since been transferred to the Linux Foundation's PyTorch Foundation, but the vulnerability remains unaddressed, leaving a large number of systems at risk of compromise and abuse for illicit cryptomining and other malicious activities.
Jun 29, 2026
CISA’s *Known Exploited Vulnerabilities (KEV) Catalog* continued to expand with newly confirmed in-the-wild exploitation, including the addition of **four CVEs**: `CVE-2019-19006` (Sangoma FreePBX improper authentication), `CVE-2021-39935` (GitLab CE/EE SSRF), `CVE-2025-40551` (SolarWinds Web Help Desk deserialization of untrusted data), and `CVE-2025-64328` (Sangoma FreePBX OS command injection). Under **BOD 22-01**, U.S. Federal Civilian Executive Branch agencies are required to remediate KEV-listed vulnerabilities by CISA’s due dates, and CISA urged non-federal organizations to use KEV as a prioritization input because these flaws are common initial access vectors. Separate reporting highlighted concerns about how CISA communicates changes to KEV metadata tied to ransomware risk: GreyNoise reported that across **59 instances in 2025**, CISA updated KEV entries to reflect **ransomware-associated exploitation** without proactively notifying defenders when the “known ransomware use” flag changed from *Unknown* to *Known*, which can materially affect patch prioritization. In parallel, third-party coverage described a CISA high-priority alert for a **critical KiloView Encoder Series** issue, `CVE-2026-1453` (CVSS **9.8**), caused by **missing authentication for critical functions** that could allow unauthenticated attackers to create/delete administrator accounts and gain full administrative control—posing disruption and lateral-movement risk in broadcast/production networks.
Jun 29, 2026