A group of major technology companies, including OpenAI and Anthropic, are forming a coalition to ensure the security of AI applications.

The Coalition for Secure AI (CoSAI) was announced at the Aspen Security Forum. This coalition was founded by several major players in the technology world, including OpenAI and Anthropic, but the group is also supported in the public market by tech giants such as AWS, Microsoft, Nvidia and Google. The goal of this coalition is to promote comprehensive protection measures to address the risks of AI. The coalition will be part of OASIS, the international standards and open source consortium.

Goals

The Coalition for Secure AI, or CoSAI for short, is the new coalition of several major players in the technology market. With this group, the companies aim to offer tools and technical guidance to help organizations secure their AI applications. In addition, it also aims to act as an ecosystem where companies can share AI-related best practices and technologies in the field of cybersecurity.

To achieve the above goals, CoSAI introduces three open source workflows, with each project handling a different subset of tasks. For example, the first project aims to help software teams find cybersecurity risks, and the second project focuses on making the work of mitigating AI cybersecurity risks easier. Finally, the third project aims to address risks in the software supply chain, which are vulnerabilities caused by software components from external sources, such as GitHub repositories.

CoSAI also focuses on developing cybersecurity risks associated with third-party AI models. Looking ahead, the coalition group plans to launch additional cybersecurity initiatives led by AI experts from the private sector and academia.