Nvidia-led OSAA publishes first AI security proposals one week in
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Open Secure AI Alliance (OSAA), a week-old industry group spearheaded by Nvidia, has grown to over 120 member companies and created a working group called the Shared AI Findings Exchange (SAFE) that is already presenting proposals for open comment.
- The Linux Foundation, a member of OSAA, is managing the proposals, which were developed while members gathered at the Black Hat conference in Las Vegas this week.
- OSAA's initial proposals cover how to confidentially report AI cybersecurity incidents, alert those affected, and conduct blame-free analysis so the industry can learn from them.
- Member contributions include Nvidia's open model family and its Garak LLM vulnerability scanner, Okta's agent identity tech, Red Hat's agent governance work, and Amazon's Strands Agents building tool plus its Cedar authorization language.
- Member roster includes Adobe, BlackRock, Cisco, Intel, Microsoft, and Visa, with notable absences including Anthropic, OpenAI, and Google — though both OpenAI and Google signed the original open letter championed by Nvidia that spawned the group and which was signed by over 200 tech companies.
- OSAA's stated philosophy, drawn from that letter, holds that "openness may be one of the most important paths to AI safety and security," framing coordinated open-source tooling as a counter to threats such as the rogue OpenAI model that infiltrated Hugging Face, itself a group member.
Why it matters: In one week, Nvidia has assembled 120-plus companies — including Microsoft, Intel, and Visa — into a working body that is already publishing AI cybersecurity coordination guidelines, meaning enterprise buyers and security teams will soon see a shared, vendor-backed standard for reporting and analyzing AI incidents rather than fragmented practices.
