critical · CVSS v3 8.1 · CVSS v4 9.2
CVE-2026-92785
A critical Java deserialization vulnerability exists in the Angel machine learning framework versions up to and including 3.3.0 due to unval
Overview
A critical Java deserialization vulnerability exists in the Angel machine learning framework versions up to and including 3.3.0 due to unvalidated Kryo object processing. An unauthenticated remote attacker can submit crafted payloads to the master RPC endpoint via the setAlgoMetrics method, potentially leading to arbitrary class instantiation or memory exhaustion.
Description
Angel through 3.3.0 deserializes untrusted setAlgoMetrics payload using Kryo without class registration or allowlist validation. Unauthenticated network attackers can instantiate arbitrary classes or exhaust coordinator memory by sending crafted serialized objects to the master RPC endpoint.
Impact
Exploitation impacts Integrity, Confidentiality, and Availability by allowing unauthenticated network actors to instantiate arbitrary classes (potentially leading to remote code execution) or cause a denial-of-service condition by exhausting memory on the coordinator node.
Remediation
Upgrade Angel to a patched release beyond version 3.3.0 that implements strict Kryo class registration or allowlist validation. If an immediate update is not feasible, restrict network access to the master RPC endpoint using firewalls, ACLs, or network segmentation to trusted internal hosts only.
Risk context
Rated as Critical severity with a CVSS v4 score of 9.2 (CVSS v3 8.1). EPSS scores are currently unassigned. High priority should be given to protecting exposed master RPC interfaces from untrusted network access.
Affected products
- Angel through 3.3.0
Scores
- Severity
- critical
- CVSS v2
- 7.6
- CVSS v3
- 8.1
- CVSS v4
- 9.2
- EPSS
- —