As artificial intelligence becomes more deeply integrated into businesses, governments and critical infrastructure, concerns over the security of open-source AI models continue to grow. To address those risks, NVIDIA and more than 35 technology companies have joined forces to launch a new industry initiative focused on improving AI security.
The companies have unveiled the Open Secure AI Alliance (OSAIA), a collaborative effort aimed at developing open standards, tools and best practices to make open-source AI systems safer and more secure. The alliance brings together major technology firms, cybersecurity companies, cloud providers and AI developers to tackle emerging security challenges as AI adoption accelerates.
Among the founding members are NVIDIA, AMD, Cisco, Dell Technologies, Google Cloud, Intel, Microsoft, OWASP Foundation, Red Hat, Trend Micro, Canonical, Sonatype and several other organizations involved in AI development and cybersecurity.
The alliance will focus on securing every stage of the AI lifecycle, from model development and training to deployment and ongoing maintenance. Members plan to build open-source security tools, establish common security frameworks and share threat intelligence to help developers identify and fix vulnerabilities before they can be exploited.
One of the group’s primary objectives is improving the security of open-source AI models, which have become increasingly popular because they allow developers to inspect, modify and build upon existing models. While openness has accelerated innovation, it has also introduced new security risks, including malicious model modifications, supply chain attacks, data poisoning and prompt injection attacks.
Rather than leaving individual organizations to solve these challenges independently, the alliance aims to create community-driven security standards that can be adopted across the AI industry. Members will also work together on security testing, vulnerability disclosure and the development of tools that help organizations assess the safety of AI models before deploying them.
The initiative comes at a time when governments and regulators are placing greater emphasis on AI security. As generative AI systems are increasingly used in healthcare, finance, software development and public services, ensuring that models are resistant to cyberattacks has become a growing priority for both technology companies and policymakers.
NVIDIA said collaboration will be essential as AI systems become more capable and more widely deployed. By bringing together companies from across the technology ecosystem, the alliance hopes to accelerate the development of practical security solutions that benefit the wider AI community instead of individual vendors.
The Open Secure AI Alliance also reflects a broader industry trend toward collaborative AI governance. In recent months, technology companies have launched several joint initiatives focused on AI safety, transparency and responsible development, recognising that many of the risks associated with advanced AI systems cannot be addressed by a single company alone.
Although the alliance is still in its early stages, its formation signals growing recognition that open-source AI security requires industry-wide cooperation. If successful, the initiative could help establish common security practices that make open-source AI models more trustworthy while supporting continued innovation across the rapidly evolving AI ecosystem.