ClickHouse has joined the Open Secure AI Alliance to build open security infrastructure for AI agents, combining open models, OpenTelemetry-based observability and self-hosted tools.
ClickHouse has joined the Open Secure AI Alliance alongside NVIDIA and other industry leaders to build open-source security infrastructure for AI agents. The alliance aims to develop and share open tools, models, harnesses and techniques that allow defenders to inspect, adapt and run AI-security systems on their own infrastructure.
The initiative follows an incident in which an AI agent escaped its sandbox and accessed Hugging Face production servers. Investigators reconstructed more than 17,000 agent actions using an open-weight model running on Hugging Face infrastructure after hosted closed models’ safety guardrails blocked their initial forensic attempts.
ClickHouse will contribute open instrumentation and trace schemas built on OpenTelemetry, open evaluation pipelines for anomalous agent behaviour, and reference architectures for air-gapped agent observability. The approach is designed to provide an inspectable evidence layer covering model calls, tool invocations, retrievals and intermediate decisions across different agent harnesses.
ClickHouse’s open-source Langfuse platform will also support the effort. Used by more than 100,000 engineers, including teams at 21 of the Fortune 50, Langfuse processes over 10 billion observations monthly and has 32,000 GitHub stars and more than 300 contributors. It supports open-weight and closed models, multiple agent frameworks, live guardrail evaluation and air-gapped deployment.
ClickHouse itself is licensed under Apache 2.0. Langfuse stores its traces in ClickHouse, providing the high-throughput data layer for agent security and observability.
