# CVE-2026-90614: FedML-AI FedML unsafe deserialization in S3 backend

Source: https://junglewise.ai/threats/cve-2026-90614-fedml-ai-fedml-unsafe-deserialization-in-s3-backend  
Publisher: Junglewise Threat Intelligence, https://junglewise.ai/threats  
Published: 2026-09-14. Updated: 2026-09-26  
License: CC BY 4.0 with a link back

CVE-2026-90614 · Severity: medium · CVSS 6.3 · Published 2026-09-14

FedML is a machine learning library used for distributed training and federated learning. A vulnerability in its S3 storage backend allows remote attackers to execute arbitrary code by sending malicious serialized objects, potentially compromising systems using this library for model training or serving.

## Quick answers

**Is CVE-2026-90614 exploited in the wild?** No. Junglewise Threat Intelligence has no report of CVE-2026-90614 being exploited in the wild.

**How severe is CVE-2026-90614?** Medium severity, with a CVSS score of 6.3, published 14 September 2026.

**Which products and versions are affected?** FedML-AI FedML up to 0.9.6.

**Is there a fix for CVE-2026-90614?** 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-2026-90614 attacked?** It is attacked over the network, in a software library.

**Am I affected by CVE-2026-90614?** Check whether anything you run is named in the advisory. 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

An unsafe deserialization vulnerability exists in the S3Storage.read_model() function within the MQTT+S3 Communication Backend (fedml/core/distributed/communication/s3/remote_storage.py). The function improperly deserializes the s3_key_str argument without validation. An attacker with network access can exploit this to craft malicious serialized payloads that execute arbitrary code upon deserialization. No authentication is required, making remote exploitation straightforward. The vendor was notified early but has not responded; patch status is unknown.

## Affected products

- FedML-AI FedML up to 0.9.6

## Timeline

- 2026-09-14: disclosed

## References

- [https://github.com/FedML-AI/FedML/](https://github.com/FedML-AI/FedML/)
- [https://github.com/FedML-AI/FedML/issues/2267](https://github.com/FedML-AI/FedML/issues/2267)