Red Hat Tightens Controls for Enterprise AI Operations

According to AI TIMES, Red Hat launched its enterprise platform, "Red Hat AI 3.5," on September 10 and at the same time strengthened the safety validation, observability, and GPU control capabilities needed to move AI into production environments. The new version also adds more than 20 new validated models to the catalog, allowing users to assess risk and performance from the model selection stage. The key point is that it brings operational issues that arise when attaching AI to business operations and services beyond pilot projects and proof-of-concept work into a single platform. Before deployment, it supports safety evaluations for models and agents as well as regulatory compliance reporting, while in operations it provides a more transparent view of inference status and resource usage. In shared GPU environments, it protects the performance of critical workloads through tenant-level resource allocation and priority-based processing, and it also reinforces security controls over external tool calls in agent development. Through this update, Red Hat made clear its direction of helping enterprises manage AI like mission-critical infrastructure.