Liquibase introduces new AI governance capabilities in Liquibase Secure, extending enterprise control to the database layer and addressing a growing blind spot in AI strategy: ungoverned database changes made by AI agents, automation scripts, and large language models interacting directly with production data.
Today, most AI governance frameworks focus on model bias, explainability and privacy, but a greater risk occurs at the data layer: AI agents that can write or modify database queries can alter or delete production data, introduce schema drift, or corrupt AI training sets before traditional security controls ever detect them.
“A single ungoverned SQL statement from an AI agent can cause more damage than months of model drift,” said Kristyl Gomes, Head of AI Strategy and Technology Innovation at Liquibase. “Enterprises think their AI governance frameworks are protecting them, but most leave the database completely exposed. Liquibase Secure brings the same level of control and traceability to data that companies already expect from code and infrastructure.”
Liquibase Secure: Database-Layer Controls for AI Workloads
Liquibase Secure provides the automation and governance infrastructure that makes AI adoption safe, compliant, and auditable.
- Automated Policy Enforcement: Blocks destructive AI-generated changes before production across 60+ database platforms
- Role-Based Approval Enforcement: Integrates with enterprise CI/CD and access controls to ensure all database changes, including those generated by AI, are reviewed and approved prior to deployment.
- Automated Drift Detection: Identifies unauthorized schema modifications and environment inconsistencies before they affect downstream systems or model training.
- Tamper-Evident Audit Trails: Creates a verifiable record of every change for frameworks such as SOX, HIPAA, GDPR, NIST AI RMF, and the EU AI Act.
- Targeted Rollback: Reverses problematic changes in minutes instead of hours
- Schema-Level Data Lineage: Captures the full history of structural evolution, which is critical for AI model provenance and regulatory audits.
Liquibase’s observability and rollback capabilities ensure that even AI-driven changes remain explainable, reversible, and fully traceable, providing a foundation for responsible AI at scale.
Liquibase Secure also introduces new AI-powered tools that accelerate delivery while maintaining control. The AI Changelog Generator, built from Liquibase’s frontline experience supporting enterprise database teams, converts natural language descriptions into validated changelogs that align with governance policies. It helps developers move from idea to production-ready change in seconds while preserving auditability and consistency.
The Liquibase Secure Developer Extension for VS Code brings schema management, history review, and policy enforcement directly into the IDE so developers can work faster without sacrificing traceability or compliance.
Together, these capabilities show how Liquibase is using AI to enhance governance, productivity, and developer experience across the database lifecycle.
MongoDB Partnership: Eliminating the Speed vs. Control Trade-Off
Liquibase also announced a new strategic technology integration with MongoDB, the unified data platform that powers modern, data-intensive, and AI-driven applications.
MongoDB’s flexible document model is a powerful enabler for rapid iteration and experimentation in dynamic AI environments. As agility drives growth, managing and tracking evolving schemas across many projects becomes a critical governance need. Issues like inconsistent field names or untracked schema drift can quietly disrupt analytics pipelines, corrupt training data, or derail audits over time.
Liquibase Secure integrates directly with MongoDB to provide continuous governance without slowing innovation. Every collection change runs through automated policy checks. Drift detection flags unapproved updates before they spread. Structured, tamper-evident logs deliver a single source of truth for auditors and data scientists.
“MongoDB customers building AI applications need both agility and governance,” said Massimiliano Marcon, Director of Product Management, MongoDB. “Our integration enables teams to experiment freely with data models while Liquibase Secure enforces consistency, prevents drift, and maintains the audit trails that emerging AI regulations demand. This integration helps enterprises move from AI experiments to production-scale AI systems.”
Regulatory Pressure
Emerging regulations demand database-layer governance. The EU AI Act requires rigorous data traceability for high-risk AI systems. NIST’s AI Risk Management Framework establishes federal and private sector baselines. Traditional frameworks, SOX, HIPAA, PCI DSS, GDPR, and DORA now intersect with AI workloads, creating compound compliance obligations.
Without database-layer controls, organizations face higher compliance costs, extended audits, and increased exposure to AI-amplified data errors.
Liquibase Secure transforms databases into AI-ready systems that balance speed, safety, and compliance. By governing schema changes across platforms such as MongoDB, PostgreSQL, Snowflake, and Databricks, Liquibase helps enterprises accelerate delivery while maintaining the trust their AI initiatives depend on.
“AI doesn’t just inherit data problems. It amplifies them faster than most teams can respond,” added Gomes. “With Liquibase Secure, organizations can finally bring the same rigor and accountability to their data that they apply to their models.”
The 2025 State of Database DevOps Report reveals that 78% of organizations struggle with AI-driven data challenges, while Gartner estimates that 40% of agentic AI projects will be canceled by 2027 if they lack clear governance at the data layer. AI governance that stops at the model is incomplete.
