NVIDIA Leads 37-Member Open Secure AI Alliance, Open-Sources NOOA Security Framework

After recent attacks on OpenAI and Hugging Face, NVIDIA formed a new group. This coalition includes Microsoft, IBM, SpaceX, and many others. They plan to standardize security measures for artificial intelligence systems. An open-source framework called NOOA will support this effort. The fallout from the cyberattacks spurred NVIDIA’s action.

A Coalition Forms Around Open Security Standards

According to its blog post and an article on The Hacker News, NVIDIA recently announced the creation of the Open Secure AI Alliance, a group with 37 members. This industry alliance focuses specifically on AI safety and security. CNBC and Decrypt report that Microsoft, IBM, and SpaceX are among the participants; this brings together companies involved in cloud infrastructure, enterprise software, and satellite operations to discuss shared security concerns.

NVIDIA has open-sourced a framework called NOOA, which The Hacker News reports is a key deliverable for the alliance; technical specifics about NOOA were not detailed in reporting. Releasing it openly instead of keeping it proprietary suggests NVIDIA intends to establish an industry standard rather than offer a product with restrictions.

Because NVIDIA is the leading provider of chips used for artificial intelligence work, its security approach has significant impact. This matters; their chips power many major AI systems. The Hill and Reuters recognize this influence. A framework from such a large supplier naturally holds more weight than one coming from a smaller company.

In plain terms

NVIDIA got 37 companies, including Microsoft, IBM and SpaceX, to agree on shared rules for keeping AI systems safe from attacks, and it's giving away the technical framework behind those rules for free rather than selling it. Because NVIDIA makes the chips almost everyone's AI runs on, when it pushes a standard, the industry tends to listen.

The Backdrop: A Rough Stretch for AI Infrastructure Security

CNBC reported that the launch timing connects directly to fallout from an OpenAI cyberattack; Reuters associates the alliance with a separate Hugging Face hack. These reports do not provide specifics regarding either incident's technical aspects, however, it’s evident that two key AI infrastructure components, a major model provider and the top platform for open models, have recently experienced attacks.

Located near major intersections, this foundation model company’s developers draw weights and datasets from repository ecosystems, encompassing a wide range of the AI supply chain. Vulnerabilities appearing at both ends of this process create considerable pressure for coordinated defenses instead of individual company efforts.

The guest list provides further insight. SpaceX’s inclusion, along with Microsoft and IBM, indicates this alliance extends beyond laboratory model security; it concerns the operational security of AI systems used within critical infrastructure and defense-adjacent workloads. A security breach in these areas carries costs exceeding mere reputational harm.

In plain terms

This alliance didn't come out of nowhere. It follows a cyberattack on OpenAI and a hack at Hugging Face, the site where most of the world's open AI models live, so the companies involved decided it made more sense to build shared defenses together than to each patch their own systems alone.

What Open-Sourcing NOOA Signals

Open-sourcing a security framework is a specific kind of move. It invites scrutiny of the framework's own design, which is generally healthier for security tooling than a closed, vendor-controlled specification that outside researchers can't audit. It also lowers the barrier for smaller companies and independent developers to adopt the same baseline without licensing friction, which is presumably part of why NVIDIA is positioning this as an industry-wide alliance rather than an internal NVIDIA initiative.

Available reports don’t specify what NOOA covers; for example, whether it considers model weight integrity or supply chain provenance for training data. Inference-time attack detection is another area that might be included. The name and timing suggest this will serve as a shared reference point, something the 37 members, and likely others not in the alliance, can use to guide their work.

Companies have mostly built their AI systems individually, each handling their own model hosting, API gateways, and training pipelines. It’s a move toward common, open frameworks based on the idea that shared standards are better than repeated work, particularly for basic security measures that don’t give anyone a special edge.

In plain terms

Making the security framework open-source means anyone can inspect it, poke holes in it, and use it without paying NVIDIA for the privilege. That's a deliberate choice to try to make this the industry's shared playbook rather than NVIDIA's private product.

Agents Going Out, Infrastructure Built to Receive Them

NVIDIA, Microsoft, IBM and SpaceX are working to secure systems like those used by OpenAI and Hugging Face, essentially protecting the infrastructure supporting AI models and developers. The Open Secure AI Alliance focuses on this outward flow of artificial intelligence - the models themselves, the technology that powers them, and the protective boundaries around these resources which AI agents utilize.

Agents, whether they’re browsers or tools like Claude or Cursor, or even GIGI agents on our portals, need reliable data when seeking businesses to book, buy from, or cite. Currently, these agents often encounter pages designed for people, requiring them to scrape and interpret information. hashtag.space and hashtag.org address this need. They offer a place for structured details about business offerings, hours, bookings, and payments. A #name and its geo-pinned portal provide that agent-readable information.

The system allows those agents to connect to the hashtag-network directly, searching, leaving leads, booking, buying, and claiming a presence. NVIDIA’s alliance handles securing these traveling agents; we aren’t building that layer or its security perimeter. Our MCP server offers this access while discovery rails BRON and CADE work toward being read and cited by AI engines, rather than optimized for some hidden algorithm. An open on-chain auction utilizes $SPACE to rank content. We aim to provide a clear, transactable space for these agents to arrive.

Claim your #name The Agentic Internet doc