Nvidia-Led Open Secure AI Alliance Launches SAFE Working Group for AI Cybersecurity

The newly formed Open Secure AI Alliance (OSAA), spearheaded by Nvidia, has launched a working group named the Shared AI Findings Exchange (SAFE) just one week after its establishment. Managed by the Linux Foundation, the coalition has rapidly expanded to over 120 member companies dedicated to enhancing artificial intelligence cybersecurity through open standards and shared technological resources.

What Happened

Gathering alongside the Black Hat cybersecurity conference in Las Vegas, members of the Open Secure AI Alliance introduced initial proposals for public comment. The framework, managed by the Linux Foundation, sets out guidelines on how to confidentially report AI cybersecurity incidents, alert impacted parties, and perform blame-free post-incident analyses to help the broader ecosystem learn from security events.

Alongside procedural standards, alliance members are cataloging open-source technologies to help organizations protect their AI systems. This initiative originated from an open letter championed by Nvidia and signed by over 200 tech companies urging the White House to support open-source AI efforts, following reports that the Trump administration was considering restrictions on Chinese open-weight models.

Key Highlights

• Rapid Growth: Spearheaded by Nvidia, the Open Secure AI Alliance expanded to more than 120 corporate members in its first week.
• SAFE Working Group: Managed by the Linux Foundation, the Shared AI Findings Exchange (SAFE) has opened initial cybersecurity guidelines for public comment.
• Standardized Reporting: Proposals outline processes for confidential incident reporting, alert notifications, and blame-free security reviews.
• Open Source Contributions: Member contributions include Nvidia’s Garak LLM vulnerability scanner and open models, Amazon’s Strands Agents building tool and Cedar authorization language, Okta’s agent identity technology, and Red Hat’s agent governance technology.
• Industry Participation: Participants include Adobe, Amazon, BlackRock, Cisco, Hugging Face, Intel, Microsoft, Okta, Red Hat, and Visa. Anthropic, OpenAI, and Google are currently absent from the alliance, although OpenAI and Google signed the original open letter.

Why This Matters

As artificial intelligence agents and models are increasingly deployed across enterprise environments, defending against rogue AI threats and vulnerabilities has become a key concern. Proponents of open-source development, including the co-founder and chief technology officer of U.S. open-weight AI lab Arcee, maintain that open collaboration and transparency serve as one of the most effective methods to address safety risks and technical threats to AI infrastructure.

What to Watch Next

The Linux Foundation is currently hosting the SAFE proposals for public comment. As the working group develops momentum, industry watchers will monitor whether major AI developers that signed the initial open letter—such as Google and OpenAI—choose to formally join the alliance.

Frequently Asked Questions

What is the Open Secure AI Alliance?

The Open Secure AI Alliance (OSAA) is an industry group spearheaded by Nvidia and supported by over 120 companies to advance AI security through shared technical tools and open standards.

What is the SAFE working group?

The Shared AI Findings Exchange (SAFE) is a working group created within the alliance and managed by the Linux Foundation to draft guidelines for reporting, alerting, and analyzing AI cybersecurity incidents.

Which companies have joined the initiative?

Members include Adobe, Amazon, BlackRock, Cisco, Hugging Face, Intel, Microsoft, Nvidia, Okta, Red Hat, and Visa, among others.

Source: Based on reporting from TechCrunch.