The vulnerability is found in MLflow’s dataset digest computation module, where a weak hash algorithm is used during digest calculation. Attackers can exploit this weakness to tamper with dataset digests, potentially injecting altered data or bypassing integrity checks. This flaw does not grant direct remote code execution but can result in unauthorized modification or fraudulent content being accepted as valid data.

MLflow, any deployment using versions up to and including 3.10.0. The fault resides in the mlflow.data.digest_utils component and affects how data sets are hashed for integrity verification.

The CVSS score of 2 indicates a low overall severity, yet the exploitability is classified as difficult, and an exploit has already been published. Attack execution is considered possible only on the local host; the attack vector appears to be local and requires high complexity. The vulnerability is not listed in the CISA KEV catalog, and no EPSS score is available, suggesting limited current exploitation activity. The risk is therefore moderate, primarily affecting local deployment security and the integrity of stored or processed data.