NVIDIA is contributing open models, weights, data, and research to the newly formed Open Secure AI Alliance, including the NVIDIA Labs Object-Oriented Agents (NOOA) framework. This initiative aims to advance AI security through open collaboration with industry leaders.
SynthePulse Insight · AI deep reading
Version 1 · 1 source
NVIDIA announced it is contributing open models, weights, data, and research to the Open Secure AI Alliance, with the core being the NVIDIA Labs Object-Oriented Agents (NOOA) framework. The framework achieves state-of-the-art results on software engineering, cybersecurity, and reasoning benchmarks, providing developers with an open foundation for optimizing agent harnesses.
On July 27, 2026, NVIDIA AI announced it is contributing open models, weights, data, and research to the newly formed Open Secure AI Alliance. The alliance, co-founded by industry leaders, aims to develop new technologies and tools for protecting software and agents. NVIDIA's contribution includes the NVIDIA Labs Object-Oriented Agents (NOOA) framework developed by its research team.
NVIDIA emphasizes that open sharing of models, tools, and research can expand the coverage of AI safety protections. This move marks a significant step for NVIDIA from internal R&D to industry collaboration in the AI safety domain.
NOOA is an object-oriented agent harness framework designed to provide a standardized runtime environment for AI agents. NVIDIA claims the framework achieves state-of-the-art results on software engineering, cybersecurity, and reasoning benchmarks. This means NOOA surpasses existing methods in several key areas, setting a new benchmark for agent performance.
As part of the open research, NOOA's code, weights, and data will be publicly available. Developers can use NOOA's findings to optimize their own harnesses and are encouraged to contribute improvements back to the community, creating a virtuous cycle.
The Open Secure AI Alliance was established to pool industry efforts to address AI safety challenges. NVIDIA's contribution includes not only the NOOA framework but also broader open models and datasets. This open model helps accelerate technological innovation and reduce redundant R&D costs.
In its announcement, NVIDIA stated that progress in AI safety requires open collaboration across the industry. By sharing research results, alliance members can more quickly identify and fix vulnerabilities, enhancing the overall security of AI systems.
This report is based on an announcement from NVIDIA AI's official X account, a primary source. The performance claims for NOOA ('achieves state-of-the-art results') are self-reported by NVIDIA and have not been independently verified. The membership composition and specific operational details of the Open Secure AI Alliance were not disclosed in the announcement.
By contributing the NOOA framework and joining the Open Secure AI Alliance, NVIDIA is advancing open collaboration in AI safety. NOOA's leading performance on multiple benchmarks suggests that object-oriented agent harness design may become a key path to improving AI performance and safety. Developers should pay attention to this framework and actively participate in community contributions.
Primary report
Primary source