This announcement follows a year of rapid innovation around Red Hat AI, including the introduction of Red Hat AI Inference Server and the launch of Red Hat AI 3. Customers around the world are using Red Hat AI across various sectors to innovate with generative, predictive, and agentic AI applications.
As organizations scale from experiments to production, they face the complex challenge of deploying models that are not only powerful but also demonstrably reliable and secure. Safety and guardrail functionality thus become a fundamental requirement for modern machine learning operations (MLOps). This focus on security and trust underscores the commitment of Red Hat and IBM to help customers with a security-first approach to responsibly scale AI in hybrid cloud environments.
By integrating Chatterbox Labs technology, a unified platform is created where safety is built in from the start. This allows Red Hat to better support organizations in bringing AI workloads into production with any model, on any accelerator, and in any environment.
Addressing unintended consequences of AI
Chatterbox Labs was founded in 2011 and brings essential technology and expertise in AI safety and transparency. The company's experience with quantitative AI risks is recognized worldwide by independent think tanks and policymakers. With this acquisition, Red Hat adds significant machine learning technology to its portfolio.
Chatterbox Labs provides automated, customized AI security and safety testing with the actual risk metrics that decision-makers need to approve AI applications for production. The technology offers a robust, model-agnostic approach to validating data and models, including through:
- AIMI for gen AI, which provides independent, quantitative risk metrics for large language models (LLMs);
- AIMI for predictive AI, which tests any AI architecture against key pillars such as robustness, fairness, and explainability;
- Guardrails that identify and mitigate unsafe, toxic, or biased prompts before models are put into production.
Securing the next generation of AI workloads
The acquisition aligns seamlessly with Red Hat's vision to support diverse models and implementation scenarios in the hybrid cloud. Furthermore, it enhances the advanced capabilities introduced with Red Hat AI 3, particularly in the area of agentic AI and Model Context Protocol (MCP).
As organizations embrace agentic AI, the importance of trusted and secure models only grows. AI agents are taking on increasingly complex, autonomous roles and can have direct impacts on critical systems.
Chatterbox Labs has conducted extensive research on the holistic security of agents, including monitoring responses and detecting MCP server actions triggered by agents. This is fully in line with Red Hat's roadmap for Llama Stack and MCP support. Red Hat is thus well-positioned to secure the next generation of intelligent, automated workloads on a reliable, enterprise-grade foundation.
By combining Red Hat's MLOps capabilities with Chatterbox Labs' guardrails, Red Hat enables organizations to operationalize their AI investments with greater confidence.
'Organizations are rapidly moving AI from the lab to production, increasing the urgency for reliable, safe, and transparent AI implementations. The innovative, model-agnostic safety testing and guardrail technology from Chatterbox Labs provide exactly the crucial 'security for AI' layer that the market needs. By integrating Chatterbox Labs into the Red Hat AI portfolio, we strengthen our promise of a complete, open source platform that enables our customers to run any model in any environment with safety built in from the start. This acquisition helps make truly responsible, production-ready AI at scale possible,' says Steven Huels, vice president, AI Engineering and Product Strategy at Red Hat.
Stuart Battersby, Ph.D., co-founder and Chief Technology Officer, Chatterbox Labs
'As AI systems permeate every aspect of business operations and the broader society, safety must not become a 'proprietary black box.' AI guardrails should not just be rolled out; they must be rigorously tested with support from demonstrable metrics. Chatterbox Labs is a pioneer in this field, from the early days of predictive AI to the agentic systems of the future. By joining Red Hat, we can bring these validated, independent safety metrics to the open-source community. The resulting transparency allows organizations to verify safety without being locked into a single vendor, thus realizing a future where we can all benefit from safe, scalable, and open AI.'
Read more about the acquisition of Chatterbox in this (English) FAQ.
