H2O.ai has joined NVIDIA-backed Open Secure AI Alliance to advance open security tools for AI agents, extending open-source principles beyond models to the wider agent stack.
H2O.ai has joined the Open Secure AI Alliance, an industry initiative led by NVIDIA and other technology companies to develop and share open technologies, techniques and tools for securing software and AI agents.
The move extends the open-source conversation beyond AI models and weights to security controls across the wider agent stack. H2O.ai also supports the Open Weights and American AI Leadership letter, which highlights open models as important for AI safety, cybersecurity, innovation and AI sovereignty.
Open weights give organisations greater visibility into deployed models and the freedom to adapt and run them on their own infrastructure. Open security tools can extend this control to agent identity, permissions, guardrails, evaluation and auditability.
As AI agents gain access to sensitive data, applications and critical workflows, H2O.ai says security must address the full agent stack, rather than whether a model is simply open or closed. Through the alliance, it will work with NVIDIA and other members to advance open approaches that help organisations trust, govern and control AI systems.
“Open source is about freedom, not just free. Sovereign AI is only possible with open weights and distributed shared philosophy of innovation,” said Sri Ambati, Founder and CEO, H2O.ai.
H2O.ai says its open-source technology is used by more than 20,000 organisations, including over half of the Fortune 500. Its open models include H2O-Danube and H2O-Mississippi, with contributions made every year since inception.
