IBM unveils watsonx.governance 2.1 to transform your AI governance experience

As organizations integrate artificial intelligence (AI) and machine learning (ML) models into their operations to drive scalability and enhance productivity, a critical question arises: how do we ensure AI workflows are responsible, secure and trustworthy?

While the promise of AI is massive it also comes with risks such as bias, lack of transparency, security vulnerabilities and compliance challenges. IBM watsonx.governance™facilitates enterprise adoption of AI to modernize business operations with a powerful AI governance, risk and compliance (GRC) tool kit built to automate AI lifecycle workflows, proactively detect and mitigate risk, and meet the changing legal, compliance and regulatory landscape.

Through effective AI governance, enterprises can assess the reliability of their AI solutions and capitalize on modern day AI tools for improved productivity and overall optimization of business costs.

Today, IBM announces the general availability of watsonx.governance 2.1 feature releases to accelerate responsible, transparent and ethical AI workflows. Here are some of the new innovations to watch out for.

IBM and AWS plan to deliver a new streamlined user experience in December that directly integrates the watsonx.governance console with Amazon SageMaker, offering customizable risk assessment and model approval workflow. The enhanced integrations will make it easier for customers to share vital information like SageMaker Model Cards and Model Registry data directly, creating a unified platform for governing AI operations. Key capabilities included are:

• Streamlined experience: Uninterrupted experience with direct integration from watsonx.governance governance console to the SageMaker model registry.
• Automatic model use case integration: Leverage integration support between an AI use case in watsonx.governance and a model group created in AWS SageMaker to reduce manual efforts of use case creation. This ensures whenever a model is created in SageMaker, it is automatically linked back to the original AI use case saving developer time and effort.
• Centralized management/unified access: Easily oversee multiple SageMaker accounts in a single instance of watsonx.governance providing a consolidated view.

This integration is currently available via the watsonx.governance SaaS offering on AWS Marketplace (AWS Marketplace).

In addition, watsonx.governance is also available as software to be deployed in a customer managed environment on both AWS and Azure with licenses available on both marketplaces.

IBM Guardium® AI Security has successfully integrated with watsonx.governance to give organizations a way to govern, secure and monitor AI in one unified experience. Some of the key capabilities include:

• Discovery of AI deployments for AI inventory: Discover AI models, data, and applications used by an organization across AI services and enrich watsonx.governance use case. Automatically identify unregistered/shadow AI deployments and trigger the appropriate action.
• Security posture insights for model risk governance: In our upcoming releases you will also be able to gain visibility into security vulnerabilities, misconfigurations and risk metrics. Unify security policy creation with risk, compliance and security stakeholders.

In today’s dynamic AI landscape driven by rapid innovation, speed is key to maintaining a competitive edge. Developers are under pressure to rapidly experiment and provide faster time to production for AI initiatives at scale.

Enter watsonx.goveranance’s Evaluation Studio that helps automate the process of reviewing individual AI assets simultaneously. Here are some of the key capabilities to watch out for, available on SaaS in Q4:  

• Faster time to production: Save developer time and effort by evaluating multiple AI assets simultaneously in one single view, paving the way for rapid experimentation.
• Improved scalability: Access actionable dashboards that help perform assessments and results against quantitative criteria, helping enterprises select the best asset for every specific AI use case and evaluate multiple AI assets efficiently at scale.
• Improved model fine tuning/calibration: In the upcoming releases you will be able to collaborate with SMEs to bring human feedback in the loop in AI/ML models. This also helps in the ability to track what is working to fine tune the model faster to get the desired results.

As AI becomes more integrated into decision-making processes across industries, enterprise exposure to potential risks for AI misuse, biased outcomes, and brand reputation have also increased. Without proper guardrails, AI could perpetuate harmful stereotypes, make biased decisions, or even cause safety hazards especially in regulated industries such as healthcare and finance.

Businesses will need to ensure that they safeguard their AI initiatives by upholding ethical AI standards and proactively mitigating potential emerging generative AI risks. watsonx.governance Guardrails helps users protect AI applications based on responsible AI practices. Explore the key capabilities:

• Ensure real-time detection: Enable real-time detection in prompt input/output with LLMs outside watsonx.ai. Handle policies on user defined tolerance thresholds.
• Mitigate emerging generative AI risks: Detect potential generative AI safety risks specifically for content—hate abuse and profanity (HAP), personally identifiable information (PII), faithfulness, answer relevance, and context relevance on LLMs in watsonx.ai as well as any third-party models.

Subsequent releases will bring additional dimensions to capture and the ability to define and manage your AI guardrails as you scale.

IBM’s watsonx.governance ensures that AI systems conform to AI regulation. It provides a clear assessment of the type of risk and an out of the box ability to assess its applicability against various AI regulations such as the EU AI Act.

In this release, we are expanding our inventory of risks to include a new category of non-technical risks. Additional description for each risk and updated names and content have been added to provide enterprises with a more holistic view of potential AI risks and its mitigation strategy.

In addition, we added new assessments for model onboarding risk identification and use case risk identification to enable proactive risk management.

Recently, the Australian government introduced new regulations for safe and responsible AI to develop a regulatory environment that builds community trust and promotes AI adoption at scale. To meet this strong demand for scalable, safe and responsible AI, IBM has launched watsonx.governance on IBM Cloud in Australia and New Zealand, significantly enhancing cloud infrastructure in the region. Now access features such as model inventory, documentation, evaluation, and monitoring through the lite and essentials plans available on IBM Cloud via the IBM Cloud catalog. Look out for more updates as we plan to expand further with more data centers globally in 2025.

In an era of growing complexity and the constant demand for AI innovation, businesses need solutions that automate AI operations, simplify the complexity of growing AI landscape, and enable seamless user experiences. IBM watsonx.governance rises to the occasion by empowering organizations to scale AI efficiently, unlocking growth and delivering meaningful business outcomes while ensuring safe and responsible AI adoption.

Try watsonx,governance to explore these new feature releases and several other enhancements built to help enterprises unlock the true potential of AI  and transform your AI governance experience today.

Try watsonx.governance for free today.
Learn how to work with modern day AI governance tools.
Learn more about watsonx.governance capabilities.