Nvidia-Led OSAA Alliance Presents Initial Results on AI Safety

The Open Secure AI Alliance (OSAA), an industry group established under Nvidia’s leadership, has achieved significant progress in just one week and brought together more than 120 companies. According to ixbt.com, a dedicated working group called Shared AI Findings Exchange (SAFE) was formed within the new alliance, and its initial proposals have been submitted for discussion. Techcrunch.com reports .

The initiative was announced during the prestigious Black Hat cybersecurity conference in Las Vegas. Although the proposed guidelines and rules are not entirely revolutionary, they cover mechanisms for confidentially reporting AI security incidents, notifying affected parties, and conducting blameless analyses to learn from mistakes.

Open-Source Technologies and Contributions from Leading Companies

Alliance participants are adding components of their open-source technologies to a shared repository and actively expanding it. In the future, these efforts could become a single open-source tool that helps enterprises protect their AI agents or defend against malicious attacks. For example, Hugging Face, a member of the group, previously experienced an infiltration involving an AI model.

Today, OSAA includes some of the world’s largest technology brands and companies:

• Nvidia — with an open model family and Garak, a tool for scanning LLM vulnerabilities;
• Okta — working on agent identity technologies;
• Red Hat — developing agent governance;
• Amazon — contributing the Strands Agents open agent-building tool and the Cedar authorization language.

Notable Participants and Conspicuous Absences

Giants such as Adobe, BlackRock, Cisco, Intel, Microsoft, and Visa have also joined the alliance. However, the absence of companies such as Anthropic, OpenAI, and Google from the group has attracted attention. Interestingly, OpenAI and Google had previously signed an open letter urging the White House to support open-source AI initiatives.

The letter, proposed by Nvidia and signed by more than 200 technology companies, emerged as the possibility of banning China’s open-weight models was being discussed in the United States. Industry experts believe that the swift action of such heavyweight players is the best way to protect the American open AI ecosystem from any external threats.