The Coalition for Secure AI (CoSAI) was announced on Thursday (July 18), at the Aspen Security Forum in Colorado.

This open-source initiative, hosted by the OASIS global standards body, aims to provide practitioners and developers with the necessary guidance and tools to create secure-by-design artificial intelligence (AI) systems.

Currently, securing AI applications and services is a fragmented endeavour, with developers grappling with inconsistent and siloed guidelines. CoSAI aims to establish standardised practices that enhance AI security and build trust globally.

CoSAI is designed to foster a collaborative and sustainable ecosystem, bringing together industry leaders, academics, and other experts to share open-source methodologies, standardised frameworks, and tools.

CoSAI has garnered support from stakeholders who all recognise the need for secure AI development. Its founding premier sponsors include tech giants Google, IBM, Intel, Microsoft, NVIDIA, and PayPal as well as founding sponsors Amazon, Anthropic, Cisco, Chainguard, Cohere, GenLab, OpenAI, and Wiz.

The initiative’s scope covers the secure building, integrating, deploying, and operating of AI systems. It focuses on mitigating risks such as model theft, data poisoning, prompt injection, scaled abuse, and inference attacks. By developing comprehensive security measures, CoSAI aims to address both classical and unique risks associated with AI systems.

CoSAI is managed by a project governing board, which advances its overall technical agenda, and a technical steering committee of AI experts from academia and industry overseeing its workstreams.

David LaBianca, Google’s CoSAI Governing Board co-chair said, “CoSAI’s establishment was rooted in the necessity of democratising the knowledge and advancements essential for the secure integration and deployment of AI. With the help of OASIS Open, we’re looking forward to continuing this work and collaboration among leading companies, experts, and academia.”