Deserialization Flaw in HanLP Model Config Loading

Executive Summary

CVE-2026-112081 is an insecure deserialization vulnerability in the HanLP library (version 2.1.0-beta.64). The issue occurs in the hanlp.load() function, which improperly resolves and instantiates classes from a model package's JSON configuration without proper validation. The vulnerability allows an attacker to craft a malicious model package that, when loaded by a victim, can instantiate any installed class compatible with the loading protocol using attacker-chosen arguments.

Useful 0 Not Useful 0

Detection Guidance

Check for unusual network activity or unexpected HTTP requests originating from HanLP processes. Inspect loaded model configurations for untrusted classpath entries or suspicious file paths. Review logs for deserialization-related errors in HanLP applications.

Useful 0 Not Useful 0

Impact Analysis

This vulnerability could allow an attacker to control the filepath parameter in the FastTextEmbedding class, triggering an HTTP GET request to an arbitrary URL. This may lead to information disclosure, tampering, or limited denial of service. The attacker could potentially escalate the attack if a protocol-compatible class with a more dangerous constructor exists.

Useful 0 Not Useful 0

Compliance Impact

This vulnerability could potentially violate compliance with GDPR and HIPAA by enabling unauthorized data access or tampering. The insecure deserialization flaw allows remote attackers to craft malicious model packages that trigger HTTP requests to arbitrary URLs, which may lead to data exfiltration or manipulation. This undermines data integrity and confidentiality requirements under these regulations.

Useful 0 Not Useful 0

Mitigation Strategies

Avoid loading model packages from untrusted sources. Restrict classpath resolution to a static allowlist of HanLP components. Validate resource paths to ensure they are local or from trusted origins. Verify checksums before model instantiation.

Useful 0 Not Useful 0