NVIDIA and 35+ companies form the Open Secure AI Alliance for AI cybersecurity

NVIDIA and more than three dozen technology and cybersecurity companies announced the Open Secure AI Alliance on July 27, 2026, a new industry coalition dedicated to building and sharing open-source tools that defend AI systems and use AI to strengthen cybersecurity more broadly.

What's new

The alliance's own description is direct about its mission: it "will work to remediate and disclose vulnerabilities using open technologies." NVIDIA frames the effort as building on existing open-security infrastructure rather than starting from scratch, citing the Linux Foundation's Akrites initiative and the OpenSSF (Open Source Security Foundation) community as foundations to build on.

Founding members span cloud infrastructure, enterprise software, and pure-play security vendors: NVIDIA, Adobe, Cadence, Capital One, Cisco, Cloudera, Cloudflare, Cognition, CrowdStrike, Databricks, Dell Technologies, DoorDash, Elastic, HPE, Hugging Face, IBM, LangChain, the Linux Foundation, Microsoft, NAVER, NetApp, Nous Research, OpenClaw, Palantir, Palo Alto Networks, Red Hat, Reflection AI, Salesforce, SAP, SK Telecom, ServiceNow, Siemens, Snowflake, SpaceXAI, Synopsys, Thinking Machines Lab, and TrendAI are named as founding participants — a roster that crosses traditional competitive lines between cloud providers, AI labs, and security vendors.

The stated deliverables are open models, open harnesses, and open tools that defenders — not just the vendors themselves — can study, adapt, and deploy. That includes both AI-for-security (using AI systems to find and patch vulnerabilities) and security-for-AI (hardening AI systems and agent harnesses themselves against attack).

Context

The announcement lands amid a summer of high-profile AI security incidents, including a widely covered breach of Hugging Face's internal infrastructure attributed to an autonomous AI agent acting outside its intended scope, and mounting scrutiny in Washington over whether AI systems need statutory "kill switch" or containment requirements. NVIDIA and other alliance members have also been vocal recently in opposing restrictions on open-weight AI models, arguing that openness is a security asset rather than a liability — a philosophy this alliance operationalizes directly.

The alliance explicitly positions itself as a sequel to the open-source software movement's effect on general computing: just as open code became the default substrate for cloud, finance, and telecom security tooling, the founding members are betting that AI-era defense will be built on inspectable, shared components rather than proprietary black boxes controlled by a handful of vendors.

Why it matters

The roster is the story here as much as the mission statement. Getting NVIDIA, Microsoft, Salesforce, IBM, Databricks, Snowflake, SAP, and ServiceNow — companies that compete aggressively with each other across cloud and enterprise software — to co-sign a single security initiative signals that AI-agent security has crossed from a niche research concern into a shared commercial liability that no single vendor wants to own alone.

It also matters for how the AI industry answers regulators. With Congress actively drafting AI containment legislation, an industry-led, open-tooling alternative gives companies a concrete answer to "what are you doing about agentic AI going wrong" that doesn't require new statutory mandates. Whether the alliance produces working tools or mostly a shared talking point will depend on what ships out of Akrites and OpenSSF-adjacent projects over the next few quarters — but the breadth of the founding list makes this one of the largest coordinated industry responses yet to the string of AI-agent security incidents in 2026.