CVE-2026-87820: cyberpanel AI Scanner endpoints

CISO Take

CyberPanel 2.4.3 through 2.4.5 ship AI Scanner debugging endpoints that require no authentication, letting any network-adjacent attacker pull administrator usernames, API-key prefixes, scan identifiers, and target domain lists straight out of the panel. There's no CISA KEV listing, no EPSS score, and no public exploit or Nuclei template yet, so this reads as a reconnaissance primer rather than an active campaign — but CyberPanel is a widely deployed multi-tenant hosting control panel, so exposed admin usernames and per-tenant domain metadata make credential-stuffing and targeted phishing against panel admins meaningfully easier, and leaked API-key prefixes narrow the search space for any follow-on key-abuse attempts. The fix is a vendor patch (GitHub commit 20484fd) that adds authentication to the AI Scanner debug routes — teams running CyberPanel 2.4.3-2.4.5 should update immediately and, in the interim, restrict access to the panel's admin interface via firewall/allowlist and rotate any API keys whose prefixes may have been exposed. Detection is straightforward: unauthenticated GET requests to AI Scanner debug/status endpoints in access logs are a clear indicator of scanning against this flaw.

Sources: NVD GitHub Advisory ATLAS

What is the risk?

Medium severity (CVSS 5.3, confidentiality-only impact) reflecting low exploitation cost — network-reachable, no privileges or user interaction required — offset by limited direct damage (no integrity/availability impact, no code execution). The real risk is enablement: this vulnerability is a reconnaissance and enumeration tool for attackers targeting CyberPanel-hosted multi-tenant environments rather than a direct compromise path. No KEV listing, no EPSS score, and no public PoC or scanner template currently exist, so near-term mass exploitation is unlikely but plausible given CyberPanel's install base and the trivial exploitation bar (a single unauthenticated GET request).

How does the attack unfold?

Reconnaissance

Attacker identifies an internet-facing CyberPanel 2.4.3-2.4.5 instance and sends unauthenticated requests to AI Scanner debug endpoints.

AML.T0006

Information Disclosure

Endpoints return administrator usernames, API-key prefixes, scan identifiers, and target domain metadata without any authentication check.

AML.T0087

Targeting

Attacker uses leaked usernames and domains to build a target list for credential stuffing or spear-phishing against panel administrators.

Impact

Successful follow-on attacks against enumerated admin accounts can lead to full panel takeover across multiple hosted tenants.

Reconnaissance

Attacker identifies an internet-facing CyberPanel 2.4.3-2.4.5 instance and sends unauthenticated requests to AI Scanner debug endpoints.

AML.T0006

Information Disclosure

Endpoints return administrator usernames, API-key prefixes, scan identifiers, and target domain metadata without any authentication check.

AML.T0087

Targeting

Attacker uses leaked usernames and domains to build a target list for credential stuffing or spear-phishing against panel administrators.

Impact

Successful follow-on attacks against enumerated admin accounts can lead to full panel takeover across multiple hosted tenants.

What systems are affected?

Package	Ecosystem	Vulnerable Range	Patched
cyberpanel	—	—	No patch

Do you use cyberpanel? You're affected.

How severe is it?

CVSS 3.1

5.3 / 10

EPSS

0.5%

chance of exploitation in 30 days

Higher than 43% of all CVEs

Source: EPSS v3 — FIRST.org

Exploitation Status

Exploit Possible

Exploitation: LOW

Sophistication

Trivial

Exploitation Confidence

low

○ Nuclei detection template available

Composite signal derived from CISA KEV, VulnCheck KEV, CISA SSVC, EPSS, Metasploit, Exploit-DB, trickest/cve, Nuclei templates, and inthewild.io exploitation reports.

What is the attack surface?

AV Network

AC Low

PR None

UI None

S Unchanged

C Low

I None

A None

What should I do?

1 step

1) Upgrade CyberPanel beyond 2.4.5 to the version containing commit 20484fd, which adds authentication to the AI Scanner debug endpoints. 2) Until patched, restrict network access to the CyberPanel admin interface and AI Scanner endpoints via firewall rules, VPN, or IP allowlisting. 3) Rotate any API keys whose prefixes could have been enumerated, and audit scanner logs for anomalous access. 4) Detection: monitor web server/access logs for unauthenticated requests to AI Scanner debug/status paths, especially bursts from a single source IP consistent with enumeration. 5) Review the vendor advisory (GHSA-qmwq-9cc8-x6h2) and VulnCheck writeup for the specific endpoint paths to alert on.

How is it classified?

Data Leakage Auth Bypass API Plugin AML.T0006 - Active Scanning AML.T0087 - Gather Victim Identity Information

Which compliance frameworks are affected?

This CVE is relevant to:

ISO 42001

A.6.2.4 - AI system data security

OWASP LLM Top 10

N/A - Not applicable — not an LLM-specific vulnerability

Frequently Asked Questions

What is CVE-2026-87820?

CyberPanel 2.4.3 through 2.4.5 ship AI Scanner debugging endpoints that require no authentication, letting any network-adjacent attacker pull administrator usernames, API-key prefixes, scan identifiers, and target domain lists straight out of the panel. There's no CISA KEV listing, no EPSS score, and no public exploit or Nuclei template yet, so this reads as a reconnaissance primer rather than an active campaign — but CyberPanel is a widely deployed multi-tenant hosting control panel, so exposed admin usernames and per-tenant domain metadata make credential-stuffing and targeted phishing against panel admins meaningfully easier, and leaked API-key prefixes narrow the search space for any follow-on key-abuse attempts. The fix is a vendor patch (GitHub commit 20484fd) that adds authentication to the AI Scanner debug routes — teams running CyberPanel 2.4.3-2.4.5 should update immediately and, in the interim, restrict access to the panel's admin interface via firewall/allowlist and rotate any API keys whose prefixes may have been exposed. Detection is straightforward: unauthenticated GET requests to AI Scanner debug/status endpoints in access logs are a clear indicator of scanning against this flaw.

Is CVE-2026-87820 actively exploited?

No confirmed active exploitation of CVE-2026-87820 has been reported, but organizations should still patch proactively.

How to fix CVE-2026-87820?

1) Upgrade CyberPanel beyond 2.4.5 to the version containing commit 20484fd, which adds authentication to the AI Scanner debug endpoints. 2) Until patched, restrict network access to the CyberPanel admin interface and AI Scanner endpoints via firewall rules, VPN, or IP allowlisting. 3) Rotate any API keys whose prefixes could have been enumerated, and audit scanner logs for anomalous access. 4) Detection: monitor web server/access logs for unauthenticated requests to AI Scanner debug/status paths, especially bursts from a single source IP consistent with enumeration. 5) Review the vendor advisory (GHSA-qmwq-9cc8-x6h2) and VulnCheck writeup for the specific endpoint paths to alert on.

What systems are affected by CVE-2026-87820?

This vulnerability affects the following AI/ML architecture patterns: AI-enabled security/malware scanning modules, multi-tenant hosting control planes.

What is the CVSS score for CVE-2026-87820?

CVE-2026-87820 has a CVSS v3.1 base score of 5.3 (MEDIUM). The EPSS exploitation probability is 0.52%.

What is the AI security impact?

Affected AI Architectures

AI-enabled security/malware scanning modulesmulti-tenant hosting control planes

MITRE ATLAS Techniques

AML.T0006 Active Scanning

AML.T0087 Gather Victim Identity Information

Compliance Controls Affected

ISO 42001: A.6.2.4

OWASP LLM Top 10: N/A

What are the technical details?

Original Advisory

CyberPanel versions 2.4.3 through 2.4.5 expose unauthenticated AI Scanner debugging endpoints that disclose administrator usernames, API-key prefixes, scan identifiers, target domains, and account metadata. Unauthenticated attackers can enumerate panel administrators and recent scanner activity to inventory multi-tenant installations and facilitate follow-on attacks.

Exploitation Scenario

An attacker scanning the internet for exposed CyberPanel instances sends unauthenticated GET requests to the AI Scanner's debugging endpoints. The responses return a list of administrator usernames, partial API keys, recent scan job IDs, and the domains each tenant has registered on the panel. The attacker cross-references the leaked usernames against credential-stuffing lists or crafts spear-phishing emails impersonating CyberPanel support, targeting the specific admin accounts and domains identified. In a hosting-provider context, this single unauthenticated request effectively hands the attacker a tenant directory to prioritize further attacks (credential attacks, subdomain takeover attempts, or social engineering) against the most valuable or least-defended accounts.

Weaknesses (CWE)

CWE-200 Exposure of Sensitive Information to an Unauthorized Actor Primary CWE-200 Exposure of Sensitive Information to an Unauthorized Actor

CWE-200 — Exposure of Sensitive Information to an Unauthorized Actor: The product exposes sensitive information to an actor that is not explicitly authorized to have access to that information.

[Architecture and Design] Compartmentalize the system to have "safe" areas where trust boundaries can be unambiguously drawn. Do not allow sensitive data to go outside of the trust boundary and always be careful when interfacing with a compartment outside of the safe area. Ensure that appropriate compartmentalization is built into the system design, and the compartmentalization allows for and reinforces privilege separation functionality. Architects and designers should rely on the principle of least privilege to decide the appropriate time to use privileges and the time to drop privileges.

Source: MITRE CWE corpus.