CyberPanel AI Scanner Auth Bypass: Unauthenticated Database Writes Expose Hosting Control Panels

TL;DR

CyberPanel versions before 2.4.4 contain a critical authentication bypass (CVE-2026-41473, CVSS 8.8) in their AI Scanner API endpoints that allows unauthenticated attackers to write arbitrary data to the backend database. The flaw opens the door to denial-of-service attacks via storage exhaustion, scan history corruption, and database field poisoning—directly compromising the integrity of hosting provider infrastructure.

What Happened

On April 24, 2026, security researchers disclosed CVE-2026-41473, an authentication bypass vulnerability in CyberPanel (a widely deployed hosting control panel) affecting all versions prior to 2.4.4. The vulnerability exists in the AI Scanner worker API endpoints, specifically:

• /api/ai-scanner/status-webhook
• /api/ai-scanner/callback

Both endpoints lack proper authentication checks, meaning any remote attacker can send malicious requests without credentials and write arbitrary data directly to the CyberPanel database. The disclosure was published through OffSeq Threat Radar and corroborated by BitNinja Security.

Technical Details

Attack Vector

The vulnerability is straightforward to exploit:

1. An attacker identifies a CyberPanel instance running version < 2.4.4.

2. The attacker crafts HTTP POST requests to the unprotected endpoints.

3. No API key, token, or authentication is required.

4. The attacker submits malicious JSON payloads that write directly to database fields.

Impact Classes

Denial of Service (Storage Exhaustion)

Attackers can flood the database with large payloads via repeated requests to status-webhook and callback, exhausting storage quota and crashing the control panel.

Scan History Corruption

The AI Scanner component stores vulnerability scan results in the database. Poisoning these records could:

• Hide real vulnerabilities from operators.
• Inject false positives to create noise and alert fatigue.
• Corrupt audit trails.

Database Field Poisoning

Attackers can overwrite critical database fields with malicious data, including user records, domain configs, or service metadata. A compromised field could serve as a pivot point for further lateral movement within the control panel.

Attack Scenario

A malicious actor could:

1. Probe for CyberPanel instances using Shodan or open port scans.

2. Automatically send requests to /api/ai-scanner/callback with payloads designed to:

- Fill storage quotas (DoS)

- Inject JavaScript or SQL payloads into ScanHistory records (for later exploitation)

- Corrupt service metadata to trigger cascading failures

Silent & Fast: The attack leaves minimal forensic traces and requires no privileged access—perfect for supply-chain attacks targeting hosting providers.

Lyrie Assessment: Why This Matters for Autonomous Defense

CyberPanel is ubiquitous in shared hosting and managed WordPress hosting environments. This vulnerability sits at the intersection of hosting infrastructure and AI tooling—a critical attack surface that Lyrie's research team identified in April as a top-tier risk.

The Autonomous Defense Angle

Traditional hosting providers rely on static API authentication models: hardcoded keys, IP whitelisting, or OAuth tokens. All are vulnerable to compromise via insider threats, supply-chain attacks, or misconfiguration.

Lyrie's autonomous defense posture would:

1. Behavioral Detection: Monitor the /api/ai-scanner endpoints for anomalous payload patterns (excessive size, unusual field writes, rapid-fire requests). Machine-speed detection catches attacks in flight, before database corruption occurs.

2. Transactional Validation: Autonomous agents could validate every database write against expected schema and business logic—rejecting writes that deviate from legitimate AI Scanner operations.

3. Forensic Reconstruction: In case of compromise, Lyrie's autonomous analysis engine could reconstruct the attack sequence and identify which database records were poisoned, enabling rapid rollback.

4. Lateral Movement Prevention: By isolating the CyberPanel service and its database with zero-trust network microsegmentation, Lyrie prevents pivot-to-admin-access scenarios.

Recommended Actions

Immediate (Next 24 Hours)

• Upgrade to CyberPanel 2.4.4 or later immediately if you run a CyberPanel instance.
• Audit your CyberPanel logs for requests to /api/ai-scanner/status-webhook and /api/ai-scanner/callback from external IPs.
• If suspicious requests are found, assume database poisoning and restore from a known-good backup pre-dating the attack.

Short Term (Next 7 Days)

• Restrict network access to the /api/ai-scanner/* endpoints using firewall rules or WAF policies. These endpoints should only accept traffic from trusted AI Scanner worker IPs.
• Deploy anomaly detection on all CyberPanel API endpoints to catch similar bypasses in the future.
• Review CyberPanel database audit logs for unusual writes to ScanHistory and service metadata fields.

Long Term (Ongoing)

• Adopt autonomous defense infrastructure that validates API authentication at the transactional level, not just the transport level.
• Segment hosting control panel infrastructure from application workloads using zero-trust network policies.
• Implement immutable audit logging for all database writes—even on "backup" systems.

Sources

1. OffSeq Threat Radar: CVE-2026-41473 (radar.offseq.com), published April 24, 2026

2. BitNinja Security: CyberPanel XSS Risks (bitninja.com), published April 25, 2026

3. CyberPanel GitHub Security Advisory (vendor patch confirmed in 2.4.4)

---

Lyrie.ai Cyber Research Division