# CVE-2026-116104: Sony Neural Network Libraries code injection in NNP Converter

Source: https://junglewise.ai/threats/cve-2026-116104-a-security-flaw-has-been-discovered-in-sony-neural-network  
Publisher: Junglewise Threat Intelligence, https://junglewise.ai/threats  
Published: 2026-10-26. Updated: 2026-10-26  
License: CC BY 4.0 with a link back

CVE-2026-116104 · Severity: medium · CVSS 6.3 · Published 2026-10-26

Sony Neural Network Libraries is a machine learning framework used to build and convert neural network models. A code injection vulnerability in its NNP Converter component allows attackers to execute arbitrary commands when converting a specially crafted model file. Since the product is no longer supported, organizations using it face ongoing risk from this publicly disclosed flaw.

Vendors: [Sony](https://junglewise.ai/threats/vendors/sony).

## Quick answers

**Is CVE-2026-116104 exploited in the wild?** No. Junglewise Threat Intelligence has no report of CVE-2026-116104 being exploited in the wild.

**How severe is CVE-2026-116104?** Medium severity, with a CVSS score of 6.3, published 26 October 2026.

**Which products and versions are affected?** Sony Neural Network Libraries up to 1.39.0.

**Is there a fix for CVE-2026-116104?** No fixed version is recorded yet. Check the vendor's advisory in the references below. If you run it, Junglewise drafts a mitigation that fits your setup and applies it once you approve, with senior security engineers behind it.

**How is CVE-2026-116104 attacked?** It is attacked over the network, in a software library.

**Am I affected by CVE-2026-116104?** Check whether you run Sony Neural Network Libraries at an affected version. Junglewise does that check for you around the clock: connect your code, cloud and domains, and it matches new vulnerabilities like this one against what you actually run and tells you when one reaches you.

## Technical details

A code injection vulnerability exists in the NNP Converter (python/src/nnabla/utils/converter/utils.py) where the attacker-controlled function.type field from an NNP model file is concatenated into Python eval() calls during the nnabla_cli convert command. A duplicate-network bypass allows malicious function.type values to reach eval() despite version validation of a separate copy, leading to arbitrary command execution. The vulnerability requires a user to convert an attacker-supplied NNP model file and is triggered remotely via the conversion operation.

## Affected products

- Sony Neural Network Libraries up to 1.39.0

## Timeline

- 2026-10-26: disclosed

## References

- [https://github.com/3em0/cve_repo/blob/main/2026/Sony-NNabla-1.39.0-Code-Injection-in-NNP-Converter-('function.type'-eval()).md](https://github.com/3em0/cve_repo/blob/main/2026/Sony-NNabla-1.39.0-Code-Injection-in-NNP-Converter-('function.type'-eval()).md)
- [https://vuldb.com/cve/CVE-2026-116104](https://vuldb.com/cve/CVE-2026-116104)

## Related threats

- [Sony Neural Network Libraries code injection in encode_param](https://junglewise.ai/threats/cve-2026-116103-a-vulnerability-was-identified-in-sony-neural-network-libraries)