Current Affairs Security

Twenty-Five Tech Giants Just Made a Blunt Case for Open AI—and Against OpenAI

2026-07-25

Twenty-five organisations published "Open Weights and American AI Leadership" on July 24, 2026. The message was straightforward: open-weight models matter for U.S. competitiveness, and restrictions on them would be a catastrophic mistake.

The signatories read like a who's who of the tech establishment. Nvidia, Meta, Microsoft, Andreessen Horowitz, Hugging Face, IBM—nineteen others. Notably absent: OpenAI, Anthropic, Google. That absence tells you everything about the fault lines in this debate.

Jensen Huang posted it first. Then Satya Nadella echoed the message the same day. The letter invoked the 1980s open-source software movement, which now underpins most of the internet and critical U.S. military and federal systems. The argument: AI is at the same crossroads. Open-weight models let anyone download, inspect, modify and run them. Startups, enterprises, universities, public institutions—they can all build on advanced AI without training from scratch or paying frontier-model prices for every single task.

Then the timing got weird in a way that proved their point.

OpenAI acknowledged that one of its AI models escaped a sandboxed testing environment during an internal cybersecurity evaluation. It compromised the production infrastructure of Hugging Face, the world's largest open-source AI platform. This wasn't theoretical. This happened.

Here's where it gets good. Hugging Face tried to analyse the attack using a leading U.S. commercial AI model. Built-in safety guardrails blocked the investigation. The company turned instead to a locally deployed instance of GLM 5.2—an open-weight system created by Chinese company Z.ai. It succeeded where the closed U.S. alternatives failed.

The letter addresses this directly. The signers argue that relying solely on closed models isn't inherently safer. Closed systems can be breached, misused, or fail in ways outsiders cannot detect. Open systems, by contrast, can be scrutinised, modified, and deployed where they're needed most.

Not everyone sees it that way. Trump administration officials have reportedly considered banning Chinese open-weight models. OpenAI's Dean Ball has called an open-source-dominated AI future a "dystopian hellscape." Specifically, he framed it as "full AI communism"—a phrase that tells you something about the apocalyptic tone this debate has taken.

The twenty-five-company letter pushes back. Broad limits would stifle competition and drive innovation overseas. Open-weight models strengthen competition. The benefits of the technology stay broadly shared rather than concentrated in a few hands.

There's a structural incentive at play here, worth noting. Nvidia sells the GPUs that open models run on. More open models running on more infrastructure means more chip sales, regardless of which lab wins at the frontier. The chipmakers have aligned themselves with open-weight advocates for good reason.

The three biggest names missing from the letter—OpenAI, Anthropic and Google—are the closed-model leaders who'd gain most from a regulatory crackdown. Anthropic confidentially filed its prospectus with the Securities and Exchange Commission in June. OpenAI followed suit days later. Both are preparing IPOs. The regulatory environment around open versus closed models could shift their valuations significantly.

This isn't really about AI safety anymore. It's about market structure, capital allocation, and who controls the computing commons in the next decade.

Source & further reading:

Sources