Enhance ML Security with JFrog Artifactory & MLflow Integration

Industry research suggests 80% or more of ML models built to create new AI-powered applications fail to deploy, largely due to technical hurdles with integrating the model into existing operations.

author-image
SMEStreet Edit Desk
New Update
Yoav Landman, CTO, JFrog

Yoav Landman, CTO, JFrog

Listen to this article
0.75x 1x 1.5x
00:00 / 00:00

 JFrog Ltd., the Liquid Software company and creators of the JFrog Software Supply Chain Platform, today announced a new machine learning (ML) lifecycle integration between JFrog Artifactory and MLflow, an open source software platform originally developed by Databricks. Following native integrations released earlier this year with Qwak and Amazon SageMaker, JFrog extends their universal AI solutions, offering organizations a single system of record with Artifactory as a model registry. The new integration gives JFrog users a powerful way to build, manage and deliver ML models and generative AI (GenAI)-powered apps alongside all other software development components in a streamlined, end-to-end, DevSecOps workflow. By making each model immutable and traceable, companies can validate the security and provenance of ML models, enabling responsible AI practices.

Industry research suggests 80% or more of ML models built to create new AI-powered applications fail to deploy, largely due to technical hurdles with integrating the model into existing operations. JFrog’s integration with MLflow helps organizations overcome this by seamlessly uniting the MLflow popular open source model development solution with an organization’s mature DevOps workflows – delivering end-to-end visibility, automation, control and traceability of ML models from experimentation to production.

“For organizations to successfully embrace and deliver AI and GenAI–powered applications at scale, developers and data science teams must manage models with trust, the same way they manage all software packages,” said Yoav Landman, CTO, JFrog. “This is only possible using a universal, scalable, single system of record for all binaries that delivers versioning, lifecycle, and security controls, which our new integration with MLflow provides.”

JFrog MLOps: A single source of truth for all models

Building on its successful integrations with all major ML tools in the market, the combination of JFrog Artifactory and MLflow enables ML engineers, Python, Java, and R developers with the freedom to work with their preferred tool stack, using Artifactory as their gold-standard model registry.  JFrog’s universal, scalable platform also natively proxies Hugging Face allowing developers to always access available open source models while simultaneously detecting malicious models and enforcing license compliance. The solution also comes with the software security features and scanners provided by the JFrog Platform to maintain risk-free ML applications.

MLSecOps - Trusted and Curated models 

The JFrog Security Research team recently discovered hundreds of instances of malicious AI ML models on the public Hugging Face AI repository posing a significant risk of data breaches or attacks. This incident highlights the potential threats lurking within AI-powered systems and underscores the need for constant security vigilance and proactive cyber hygiene.

Uniting JFrog Artifactory with MLflow will empower users to more easily build, train, and deploy models with greater security, governance, versioning, traceability, and trust by leveraging  JFrog’s scanning environment to rigorously examine every new model uploaded to Hugging Face. 

For a deeper look at JFrog’s integration with MLflow to power ML and GenAI-powered app development, read this blog post. Developers interested in going hands-on with these new features can download the free plug-in here

ML Security JFrog Artifactory