CriticalCVE-2025-62593

    Ray is vulnerable to RCE via Safari & Firefox Browsers through DNS Rebinding Attack (CVE-2025-62593)

    Published Last modified In CISA KEV since Sources: pwn.watch, CVE.org, Red Hat, OSV

    Share

    What is affected, what to do

    A flaw was found in Ray’s HTTP API endpoint handling (e.g. /api/jobs, /api/job_agent/jobs/), which allows a remote attacker to trigger arbitrary code execution when a developer using Ray visits a malicious website in a vulnerable browser (e.g. Firefox or Safari).

    The vulnerability is being actively exploited; treat it as a priority. Update Ray to version 2.52.0. Immediate actions, mitigation and detection rules are in the runbook below.

    Affected
    • Ray lt *..2.52.0
    Fixed in
    • Ray 2.52.0

    Not sure whether and how you are affected? Ask a question about this vulnerability

    Metrics

    9.4

    Priority

    XIEM Effective CVSS: highest available CVSS rating from NVD and vendor data.

    Source: NVD, CVSS v4

    8.8

    CVSS v3

    Severity High on the scale from 0 to 10.

    Source: pwn.watch, CVE.org, Red Hat, OSV

    60.2 %

    EPSS score

    High — model estimates ≥ 50% chance of real-world exploitation within 30 days.

    Source: FIRST EPSS, percentile 99.14

    Show all metrics
    Severity
    Critical
    Priority (0–325)

    The priority score is a consolidated metric that condenses several individual signals into a single, action-guiding ranking: technical severity (CVSS), exploitation likelihood (EPSS), known-exploited status (KEV) and the context of the affected asset. The aim is to answer the only truly important question in vulnerability management: what first?

    Read in glossary
    237.88 Methodology
    Exploit maturity

    An exploit is code or a technique that specifically uses a concrete vulnerability to force unintended behaviour — such as executing code or escalating privileges. It is the tool that turns a theoretical flaw into a practical attack.

    Read in glossary
    actively exploitedSource: VulnCheck KEV (opens in a new tab) How we classify
    XIEM Effective CVSS

    XIEM® is the tiered managed-security operating model from NEOSEC and a registered trademark of NEOSEC GmbH. It extends classic SIEM — the central collection, normalisation and correlation of security-relevant events — with endpoint forensics, network visibility and deception technology, and consolidates them into three building tiers: Sentry, Orchestrate and Command.

    Read in glossary
    9.4 Methodology
    CVSS v3

    CVSS is the globally established, open standard for assessing the technical severity of a vulnerability on a scale from 0 to 10. It describes how easily a flaw can be exploited and what damage it can cause — but not how likely its actual exploitation is. CVSS and the complementary EPSS deliberately answer different questions.

    Read in glossary
    8.8
    CVSS v4

    CVSS is the globally established, open standard for assessing the technical severity of a vulnerability on a scale from 0 to 10. It describes how easily a flaw can be exploited and what damage it can cause — but not how likely its actual exploitation is. CVSS and the complementary EPSS deliberately answer different questions.

    Read in glossary
    9.4
    EPSS score

    The EPSS score is the absolute model output for an individual vulnerability: the estimated probability of real-world exploitation within the next 30 days, expressed as a value between 0 and 1 (i.e. 0 % to 100 %). A score of 0.42 means „the model estimates a 42 % chance of real-world exploitation within 30 days“.

    Read in glossary
    60.2 %
    High — model estimates ≥ 50% chance of real-world exploitation within 30 days.
    EPSS percentile

    The EPSS percentile is the relative position of a vulnerability within all EPSS-scored CVEs, expressed as a value between 0 and 1. A percentile of 0.90 means „this CVE has a higher EPSS score than 90 % of all other scored CVEs“. Unlike the EPSS score, it is not a probability — but a positioning within a ranking.

    Read in glossary
    99.14
    Critical — this CVE ranks in the top fifth of all CVEs scored today (rank ≥ 80%).
    KEV status

    The KEV catalogue is a list maintained by the US authority CISA of vulnerabilities for which active exploitation in the wild has been confirmed. It answers not the question „how severe?“ but „is it already being attacked?“ — thereby sharply prioritising patching.

    Read in glossary
    Actively exploited
    Published
    In CISA KEV since

    The KEV catalogue is a list maintained by the US authority CISA of vulnerabilities for which active exploitation in the wild has been confirmed. It answers not the question „how severe?“ but „is it already being attacked?“ — thereby sharply prioritising patching.

    Read in glossary
    CISA assessment (SSVC)
    Exploitation: active · automatable · Technical impact: total
    Source: CISA Vulnrichment, as of 18 August 2026
    CWE

    CWE is a catalogue of typical weakness types in software and hardware — such as flawed input validation, unsafe memory handling or poor authentication. Where a CVE names a concrete vulnerability in a concrete product, a CWE describes the underlying class of defect.

    Read in glossary
    CWE-94, CWE-352

    Response and mitigation

    Why act now?

    Prioritisation rationale

    The flaw is network-reachable without authentication and needs only a developer with a running Ray instance to open a crafted page in Firefox or Safari. Ray's Jobs API has no built-in authentication, so bypassing the User-Agent check leads directly to full code execution as that developer. Attackers favour AI/ML teams, data-science groups and research institutions, whose workstations concentrate cloud credentials, GPU access, model weights and training data. Exploitation likelihood is among the highest of any CVE, while CISA reports no known ransomware use so far. Organisations with active ML development should treat this as a high-priority upgrade within 24-72 hours; where Ray is absent, it is an inventory task only.

    1. Runbook step 1

      Immediate response (0-24 h)

      • Inventory: Find every developer workstation, laptop, Jupyter/VS Code environment and CI runner running Ray locally (pip show ray, ray --version, processes raylet, gcs_server, dashboard.py). The exposure is primarily local dev instances, not just production clusters.
      • Upgrade: Move Ray to the fixed release named in the vendor advisory (the Ray project's GitHub Security Advisory). The context references version 2.52.0. Check the advisory to confirm whether that is the fixed or an affected release, then pin it in requirements.txt/pyproject.toml/container images.
      • Until patched, disable the dashboard and Jobs API: Start Ray with ray start --include-dashboard=false or ray.init(include_dashboard=False), and ray stop idle local clusters. This removes the vulnerable /api/jobs and /api/job_agent/jobs/ endpoints (dashboard default port 8265).
      • Browser hygiene: Tell developers running Ray not to browse untrusted sites in Firefox or Safari until they have upgraded. Use a Chromium-based browser with Private Network Access protection for dev work, or stop Ray before browsing.
      • Compromise check: Review ray job list and the dashboard logs under /tmp/ray/session_*/logs/ on affected hosts for unknown job submissions. On any hit, isolate the host and rotate every credential reachable from it: cloud CLI profiles, SSH keys, Git tokens, Hugging Face and other API keys.

    2. Runbook step 2

      Mitigation layers

      • DNS layer (anti-rebinding): Configure internal resolvers to drop public-domain answers that point at loopback or RFC1918 space, e.g. Unbound private-address: 127.0.0.0/8, 10.0.0.0/8, 172.16.0.0/12, 192.168.0.0/16, or dnsmasq stop-dns-rebind. Extend this to VPN clients and remote workers.
      • Network segmentation: Never bind Ray ports (dashboard 8265, Ray Client 10001, GCS 6379) to 0.0.0.0. Host firewalls should accept inbound connections to these ports only from 127.0.0.1. Production clusters should be reachable only from a dedicated management VLAN.
      • Reverse proxy with Host validation: If the dashboard is required, front it with an authenticating nginx/Envoy proxy. The proxy rejects any request whose Host header is not exactly localhost:8265 or the expected FQDN, and it enforces an Origin check. This neutralises DNS rebinding.
      • Endpoint hardening: Use EDR or AppLocker/WDAC rules to block Ray worker processes from spawning shells and network tools (bash, sh, curl, wget, powershell.exe). Run Ray development in isolated containers or VMs that have no access to personal credentials.
      • IAM / least privilege: Don't run Ray under admin accounts or alongside long-lived cloud credentials. Use short-lived SSO/STS tokens so that code execution has limited reach.

    3. Runbook step 3

      Detection rules

      • Dashboard/proxy access logs: POST /api/jobs/ or /api/job_agent/jobs/ with a Host header other than localhost, 127.0.0.1 or the expected cluster name is a strong rebinding indicator. SPL: index=ray uri_path="/api/job*" method=POST NOT host_header IN ("localhost:8265","127.0.0.1:8265")
      • EDR process ancestry (Sysmon EID 1 / auditd execve): raylet, ray::JobSupervisor or a Ray-context python spawning sh, bash, curl, wget, nc or base64. Sigma shape: ParentCommandLine|contains: 'ray' AND Image|endswith: ['/bash','/sh','/curl','/wget']
      • DNS telemetry (Zeek dns.log, Sysmon EID 22): External domains that resolve first to a public IP and then to 127.0.0.1 or RFC1918 space, with very low TTLs. KQL: DnsEvents | where IPAddresses has_any ("127.0.0.1","0.0.0.0") and not(Name endswith ".local" or Name endswith ".internal")
      • Network connections (Sysmon EID 3): Ray worker processes connecting to unfamiliar external hosts shortly after a job starts. This typically indicates payload staging or exfiltration of ~/.aws, ~/.ssh or ~/.config.
      • Ray job metadata: Entrypoints containing curl … | bash, python -c, Base64 blobs or reverse-shell patterns. Poll ray job list --format json and forward the output to the SIEM.

    Detection rules

    No NEOSEC detection rule is available for this vulnerability yet.

    Classification under NIS2 and CRA

    Rule-based classification from exploitation status and severity. It does not replace a case-by-case assessment.

    NIS2

    NIS2 is EU Directive (EU) 2022/2555 to raise the level of cybersecurity across the Union. It substantially broadens obligations compared with the first NIS directive: more sectors, mandatory risk management, staged reporting duties and personal accountability of senior management.

    Read in glossary
    Actively exploited: treat the vulnerability as a priority in vulnerability handling under Art. 21(2)(e) NIS2. If it is exploited in your environment, check the reporting obligation under Art. 23 (early warning within 24 hours).
    CRA

    The CRA is EU Regulation (EU) 2024/2847 setting horizontal cybersecurity requirements for „products with digital elements“ — from industrial controllers and software to connected consumer devices. For the first time manufacturers must ensure security across the entire product lifecycle, handle vulnerabilities and provide security updates.

    Read in glossary
    Since 11.09.2026, manufacturers of affected products with digital elements must report actively exploited vulnerabilities under Art. 14 CRA, with an early warning within 24 hours via ENISA's single reporting platform.
    ISO/IEC 27001

    ISO/IEC 27001 is the internationally leading standard for an information security management system (ISMS). It describes a risk-based, process-oriented framework to systematically plan, implement, review and improve information security.

    Read in glossary
    Control A.8.8, management of technical vulnerabilities: assess relevance to your systems, plan remediation and keep evidence.

    Description

    A flaw was found in Ray’s HTTP API endpoint handling (e.g. /api/jobs, /api/job_agent/jobs/), which allows a remote attacker to trigger arbitrary code execution when a developer using Ray visits a malicious website in a vulnerable browser (e.g. Firefox or Safari). The root cause is an insufficient defense relying solely on the User-Agent header starting with “Mozilla”, which can be manipulated under the fetch specification — enabling a DNS-rebinding attack to bypass browser-based protections.

    Source: pwn.watch, NVD (NIST), CISA KEV Catalog, CVE.org (MITRE), redhat_csaf, cvelistv5, OSV (Open Source Vulnerabilities)

    Affected products

    The following versions are affected. Older releases of the same product line are also vulnerable unless stated otherwise.

    Affected products
    Affected versionsFixed in
    ray-projectray
    • < 2.52.0
    • 2.52.0
    anyscaleray
    • < 2.52.0
    • 2.52.0

    Vendors and products

    Weakness classes (CWE)

    • CWE-94Improper Control of Generation of Code ('Code Injection')

      Base

      The product constructs all or part of a code segment using externally-influenced input from an upstream component, but it does not neutralize or incorrectly neutralizes special elements that could modify the syntax or behavior of the intended code segment.

      Entry at MITRE CWE-94

    • CWE-352Cross-Site Request Forgery (CSRF)

      Compound

      The web application does not, or cannot, sufficiently verify whether a request was intentionally provided by the user who sent the request, which could have originated from an unauthorized actor.

      Entry at MITRE CWE-352

    Public exploit references

    Publicly reported proof-of-concepts, exploit modules and detection templates for this vulnerability. NEOSEC links to the original sources only and does not provide any exploit code.

    Public exploit references
    CISA KEV CatalogCISA Known Exploited Vulnerability (opens in a new tab)cisa.govactively exploited
    Reference tagged as exploitgithub.com (opens in a new tab)PoC (publicly reported)reported, not verified—
    PoC-in-GitHub (nomi-sec)CVE-2026-33017 (opens in a new tab)github.comPoC (publicly reported)reported, not verified
    PoC-in-GitHub (nomi-sec)CVE-2025-62593-PoC (opens in a new tab)github.comPoC (publicly reported)reported, not verified

    References & sources

    • github.comray-project/ray/security/advisories/GHSA-q279-jhrf-cc6vweb, Exploit, Patch, Vendor advisory, x_refsource_CONFIRM, pwn-watch
    • NVD (NIST)NVD: CVE-2025-62593advisory, pwn-watch
    • github.comnccgroup/singularity/pull/68web
    • github.comray-project/ray/commit/70e7c72780bdec075dba6cad1afe0832772bfe09web, Patch, x_refsource_MISC
    • docs.ray.ioen/releases-2.51.1/ray-security/index.htmlweb
    • en.wikipedia.orgwiki/Malvertisingweb
    • github.comray-project/raypackage
    • github.comray-project/ray/blob/e7889ae542bf0188610bc8b06d274cbf53790cbd/p…web
    • github.comray-project/ray/blob/f39a860436dca3ed5b9dfae84bd867ac10c84dc6/p…web
    • pypi.orgproject/raypackage
    • GitHub Security Advisoriesadvisories/GHSA-q279-jhrf-cc6vadvisory
    • bitsight.comblog/rondodox-botnet-infrastructure-analysisThird-party advisory, Third-party advisory
    • cisa.govknown-exploited-vulnerabilities-catalog?field_cve=CVE-2025-62593government-resource, US government resource
    • pwn.watchpwn.watch: Ray-Project Ray: Ray-Project Ray Code Injection Vulnerabilitypwn-watch
    • cisa.govCISA Known Exploited Vulnerabilities Catalogpwn-watch

    Changes to this CVE

    Who changed this CVE and when: NVD (analysis, affected products, CVSS), the CVE publisher (CNA) and additional providers such as CISA (ADP).

    1. Record modifiedGitHub

      Show 6 original changes
      Individual changes of this event as published by NVD
      TypeActionBeforeAfter
      Referenceadded
      —
      https://github.com/ray-project/ray/commit/70e7c72780bdec075dba6cad1afe0832772bfe09
      Referenceadded
      —
      https://github.com/ray-project/ray/security/advisories/GHSA-q279-jhrf-cc6v
      Referenceremoved
      https://github.com/ray-project/ray/commit/70e7c72780bdec075dba6cad1afe0832772bfe09
      —
      Referenceremoved
      https://github.com/ray-project/ray/security/advisories/GHSA-q279-jhrf-cc6v
      —
      Reference typeremoved
      https://github.com/ray-project/ray/commit/70e7c72780bdec075dba6cad1afe0832772bfe09 Types: Patch
      —
      Reference typeremoved
      https://github.com/ray-project/ray/security/advisories/GHSA-q279-jhrf-cc6v Types: Exploit, Patch, Vendor Advisory
      —

      Source identifier: security-advisories@github.com

    2. Record modifiedCISA ADP

      CISA assessment (SSVC): exploitation: active, automatable: yes

      Show 8 original changes
      Individual changes of this event as published by NVD
      TypeActionBeforeAfter
      Referenceadded
      —
      https://www.bitsight.com/blog/rondodox-botnet-infrastructure-analysis
      Referenceadded
      —
      https://www.cisa.gov/known-exploited-vulnerabilities-catalog?field_cve=CVE-2025-62593
      Referenceremoved
      https://www.bitsight.com/blog/rondodox-botnet-infrastructure-analysis
      —
      Referenceremoved
      https://www.cisa.gov/known-exploited-vulnerabilities-catalog?field_cve=CVE-2025-62593
      —
      Reference typeremoved
      https://www.bitsight.com/blog/rondodox-botnet-infrastructure-analysis Types: Third Party Advisory
      —
      Reference typeremoved
      https://www.cisa.gov/known-exploited-vulnerabilities-catalog?field_cve=CVE-2025-62593 Types: US Government Resource
      —
      SSVC assessmentremoved
      {"id":"CVE-2025-62593","role":"CISA Coordinator","options":[{"exploitation":"active"},{"automatable":"yes"},{"technicalImpact":"total"}],"version":"2.0.3","timestamp":"2026-08-17T00:00:00+00:00"}
      —
      SSVC assessmentadded
      —
      {"id":"CVE-2025-62593","role":"CISA Coordinator","options":[{"exploitation":"active"},{"automatable":"yes"},{"technicalImpact":"total"}],"version":"2.0.3","timestamp":"2026-08-18T03:55:34.674806Z"}

      Source identifier: 134c704f-9b21-4f2e-91b3-4a467353bcc0

    3. Record modifiedCISA ADP

      CISA assessment (SSVC): exploitation: active, automatable: yes

      Show the original change
      Individual changes of this event as published by NVD
      TypeActionBeforeAfter
      SSVC assessmentchanged
      {"id":"CVE-2025-62593","role":"CISA Coordinator","options":[{"exploitation":"active"},{"automatable":"no"},{"technicalImpact":"total"}],"version":"2.0.3","timestamp":"2026-08-17T17:40:04.986785Z"}
      {"id":"CVE-2025-62593","role":"CISA Coordinator","options":[{"exploitation":"active"},{"automatable":"yes"},{"technicalImpact":"total"}],"version":"2.0.3","timestamp":"2026-08-17T00:00:00+00:00"}

      Source identifier: 134c704f-9b21-4f2e-91b3-4a467353bcc0

    4. CISA KEV entry updated9119a7d8…

      2 individual changes: Date Added, Due Date

      Show 2 original changes
      Individual changes of this event as published by NVD
      TypeActionBeforeAfter
      Date Addedchanged
      2026-08-18
      2026-08-17
      Due Datechanged
      2026-08-18
      2026-08-17

      Source identifier: 9119a7d8-5eab-497f-8521-727c672e3725

    5. Record modifiedCISA ADP

      CISA assessment (SSVC): exploitation: active, automatable: no

      Show the original change
      Individual changes of this event as published by NVD
      TypeActionBeforeAfter
      SSVC assessmentchanged
      {"id":"CVE-2025-62593","role":"CISA Coordinator","options":[{"exploitation":"active"},{"automatable":"no"},{"technicalImpact":"total"}],"version":"2.0.3","timestamp":"2026-08-17T14:46:35.844911Z"}
      {"id":"CVE-2025-62593","role":"CISA Coordinator","options":[{"exploitation":"active"},{"automatable":"no"},{"technicalImpact":"total"}],"version":"2.0.3","timestamp":"2026-08-17T17:40:04.986785Z"}

      Source identifier: 134c704f-9b21-4f2e-91b3-4a467353bcc0

    Show 3 older changes

    1. CISA KEV entry updated9119a7d8…

      4 individual changes: Date Added, Due Date, Required Action, Vulnerability Name

      Show 4 original changes
      Individual changes of this event as published by NVD
      TypeActionBeforeAfter
      Date Addedadded
      —
      2026-08-18
      Due Dateadded
      —
      2026-08-18
      Required Actionadded
      —
      2026-08-18
      Vulnerability Nameadded
      —
      2026-08-18

      Source identifier: 9119a7d8-5eab-497f-8521-727c672e3725

    2. Record modifiedCISA ADP

      CISA assessment (SSVC): exploitation: active, automatable: no

      Show 2 original changes
      Individual changes of this event as published by NVD
      TypeActionBeforeAfter
      Referenceadded
      —
      https://www.cisa.gov/known-exploited-vulnerabilities-catalog?field_cve=CVE-2025-62593
      SSVC assessmentchanged
      {"id":"CVE-2025-62593","role":"CISA Coordinator","options":[{"exploitation":"poc"},{"automatable":"no"},{"technicalImpact":"total"}],"version":"2.0.3","timestamp":"2025-11-28T18:21:16.960626Z"}
      {"id":"CVE-2025-62593","role":"CISA Coordinator","options":[{"exploitation":"active"},{"automatable":"no"},{"technicalImpact":"total"}],"version":"2.0.3","timestamp":"2026-08-17T14:46:35.844911Z"}

      Source identifier: 134c704f-9b21-4f2e-91b3-4a467353bcc0

    3. Record modifiedCISA ADP

      Show the original change
      Individual changes of this event as published by NVD
      TypeActionBeforeAfter
      Referenceadded
      —
      https://www.bitsight.com/blog/rondodox-botnet-infrastructure-analysis

      Source identifier: 134c704f-9b21-4f2e-91b3-4a467353bcc0

    One NVD translation into another language is hidden.

    Embed badge

    Show the severity and KEV status of this vulnerability on your site – with a link back here. The image is updated hourly.

    Badge preview for CVE-2025-62593
    <a href="https://intel.neosec.eu/en/cve/CVE-2025-62593"><img src="https://intel.neosec.eu/badge/cve/CVE-2025-62593.svg?lang=en" alt="CVE-2025-62593 – NEOSEC Intel" height="20"></a>