New Model Validation Capabilities Ensure Transparency, Compliance, and Trust in AI for Financial Services and Banks

MOUNTAIN VIEW, Calif.--(BUSINESS WIRE)--H2O.ai, the leader in open-source Generative AI and the most accurate Predictive AI platforms, today announced the industry's first Model Risk Management (MRM) framework for Generative AI, bringing rigorous validation, compliance, and transparency to Generative AI applications in financial services, banking, and other highly regulated sectors.

As AI adoption accelerates, particularly in regulated industries, ensuring the trustworthiness, fairness, and reliability of Generative AI models and applications is paramount. H2O.ai’s MRM solution provides a structured evaluation framework that integrates automated testing and evaluation with human calibration, model weakness and failure identification, bias detection, and explainability tools, offering enterprises the ability to validate and mitigate AI-related risks before deployment.

Why It Matters for Regulated Industries

Financial institutions and banks operate under strict regulatory guidelines requiring model transparency, robustness, and explainability to mitigate risks like biased decision-making, hallucinated outputs, or security vulnerabilities. H2O’s Model Risk Management framework extends traditional MRM principles to Generative AI, providing:

• Automated Test Generation – Generate diverse query types using topic modeling, stratified sampling, and LLM-based test generation, with selection guided by embedding-based verification metrics.
• Embedding-Based Functionality Metrics – Measure the model’s ability to retrieve, synthesize, and generate accurate responses to user queries.
• Human-Calibrated Evaluations – Align machine evaluation with human judgment through a calibration model and conformal prediction techniques.
• Weakness Identification and Risk Mitigation – Identify areas of low performance through bivariate analysis and failure clustering, enabling targeted improvements and risk mitigation via guardrails.
• Robustness Testing – Assess model robustness with adversarial inputs, out-of-distribution queries, and input variations introduced through prompt perturbation and noise injection.
• Transparency and Explainability – Enhance transparency and explainability through ML-based evaluation, visualization tools, and interactive widgets.

"Regulated industries need trustworthy AI that meets strict compliance, risk, and transparency requirements," said Sri Ambati, CEO and founder of H2O.ai. "By bringing rigorous Model Risk Management to Generative AI, we are enabling banks and financial services to confidently deploy AI solutions with auditable, explainable, and reliable outcomes."

“The release of our latest software marks a significant leap forward in the validation and testing of generative language models, particularly in high-stakes applications like banking. Built upon the Human-Calibrated Automated Testing (HCAT) framework, this software introduces a structured, scalable, and transparent approach to evaluating Generative Language Model systems. By integrating automated test generation, embedding-based functionality metrics, and human-calibrated evaluations, we ensure that AI-driven solutions meet the highest standards of accuracy, reliability, and regulatory compliance. Our commitment to explainability, robustness, and risk mitigation empowers organizations to deploy generative AI with confidence, knowing that their models have undergone rigorous, human-aligned assessment,” said Agus Sudjianto, Senior Vice President, Risk and Technology for Enterprise at H2O.ai.

Scaling AI Expertise in Financial Services

H2O.ai has trained AI practitioners, risk teams, and model validators at leading banks, including CBA, Wells Fargo, KeyBank, USAA, US Bank, UBS, Comerica, Northern Trust, Fifth Third, MUFG, Barclays, HSBC, Ally Bank, and Discover. By equipping them to test, monitor, and validate Generative AI models, H2O.ai has helped institutions build in-house AI expertise to reduce reliance on third-party validation, and enables faster, safer, and more cost-effective AI deployment in financial services.

Available Airgapped and On-Prem for Maximum Security

H2O.ai’s Model Risk Management capabilities are now available as part of Enterprise h2oGPTe, supporting airgapped and on-premise deployments to ensure compliance with data sovereignty, security, and privacy mandates. This allows financial institutions to validate and monitor AI models securely within their own infrastructure, reducing third-party risk exposure.

H2O.ai continues to lead the industry by converging Predictive and Generative AI with enterprise-grade risk management, compliance, and automation capabilities. With this latest MRM release, organizations can now deploy validated, high-performing AI models that meet the most demanding regulatory requirements.

For more information or to schedule a demo, visit https://h2o.ai/platform/enterprise-h2ogpte/model-validation/

About H2O.ai

Founded in 2012, H2O.ai is at the forefront of the AI movement to democratize Generative AI. H2O.ai’s open-source Generative AI and Enterprise h2oGPTe, combined with Document AI and the award-winning autoML Driverless AI, have transformed more than 20,000 global organizations, and over half of the Fortune 500, including AT&T, Commonwealth Bank of Australia, Chipotle, Singtel, Workday, Progressive Insurance, and AES.

H2O.ai partners include Dell, Deloitte, Ernst & Young (EY), PricewaterhouseCoopers (PwC), NVIDIA, Snowflake, AWS, Google Cloud Platform (GCP) and Microsoft Azure. H2O.ai’s AI for Good program supports nonprofit groups, foundations, and communities in advancing education, healthcare, and environmental conservation. With a vibrant community of 2 million data scientists worldwide, H2O.ai aims to co-create valuable AI applications for all users.

H2O.ai has raised $256 million from investors, including Commonwealth Bank, Nvidia, Goldman Sachs, Wells Fargo, Capital One, Nexus Ventures and New York Life.
Contacts

Media Contact

Betty Candel
VP Marketing
betty.candel@h2o.ai