A race condition exists in Splunk AI Toolkit versions below 6.0.0 that allows an attacker who can upload models to overwrite a model being uploaded by another user. The concurrent upload request replaces the original model content, causing the model lookup entry to reference attacker‑controlled data. Because the toolkit does not verify ownership of the uploaded content, the integrity of the model repository is compromised, potentially leading to the use of malicious or altered models in downstream analytics and visualizations.

Splunk AI Toolkit in all versions prior to 6.0.0 are affected. The vulnerability applies to any installation where multiple users or services can invoke the model upload REST API, including environments that use Splunk App for Data Science and Deep Learning or custom search commands that rely on AI Toolkit models.

The CVSS score of 5.9 indicates a moderate severity, but the exploitation requires a user with model‑upload privileges and the ability to coordinate concurrent requests to the same model name. The EPSS score is not provided, and the vulnerability is not listed in CISA’s KEV catalog, suggesting a lower publicly observed exploitation probability. The likely attack vector is the HTTP REST API used for model uploads, and an attacker could locally or remotely trigger the race condition in a compromised or privileged account. If the exploit succeeds, it can subvert authenticated model use without triggering administrative alerts, leading to integrity and confidentiality impacts in downstream data processing.