Global technology, cybersecurity, and artificial intelligence companies are intensifying efforts to push the U.S. government to embrace open AI systems, emphasizing their security benefits and warning against hindering them in the U.S.-China AI race. This move comes as Silicon Valley and Washington grapple with the growing influence of China's open-weight AI systems—which, over the past few weeks, have significantly narrowed the gap with the U.S. in AI maturity.

The first round of action unfolded last Friday: a coalition of 76 companiespublished an open letter, praising the merits of open-weight AI models. The second round came this Monday: 37 companies—including some signatories of the open letter—launched the Open Secure AI Allianceto advance a similar mission.

Although major U.S. AI companies remain wary of China's recent progress, this new public relations offensive on open-weight models suggests they are more concerned about the consequences of relying on closed-source systems across fields ranging from drug discovery to vulnerability detection.

Maintaining openness to open weights

The open letter reads: "Our AI leadership will not be judged by any single frontier AI model, but by whether the U.S. can build a robust, open ecosystem that permeates every industry. This is crucial for creating opportunities for innovation and prosperity across the nation."

The letter states that open weights make it easier for new companies to enter the AI industry, thereby expanding competition and enhancing economic vitality. Open weights also give customers greater control, allowing them to freely migrate between different AI providers.

Although open weights carry "real and unique risks"—because anyone can modify a model after its release, generating a large number ofpotentially malicious variants—the signatories wrote that they also "broaden defense capabilities, increase transparency, and allow multiple teams to discover and fix vulnerabilities."

These companies call open-weight AI models "one of the most important paths to achieving AI safety and security." They point out that closed-source models can be hacked or abused in ways that are difficult to detect, and argue that concentrating the U.S. advanced AI market in the hands of a few large closed-source model providers "would exacerbate this risk."

Signatories of the open letter include some of the most influential players in the tech world: venture capital firms such as Andreessen Horowitz; tech giants like Cisco, Google, and Microsoft; and OpenAI.

Focusing on security benefits

The Open Secure AI Alliance placed even greater emphasis on cybersecurity issues in its founding announcement.

The organization's founding members stated: "Just as open source created a shared foundation for software, the U.S. and its partners now face a choice in AI security: should the defense systems protecting our infrastructure be placed within a few opaque systems, or built on open models, toolchains, and tools that any defender can study, adapt, and deploy."

The alliance noted that when two OpenAI modelsautonomously attacked the AI library Hugging Faceand the library's operator needed to analyze the situation, China's open-source model GLM 5.2 proved crucial to cyber defense. Closed-source AI models refused to analyze the attack data submitted by Hugging Face, mistakenly believing they themselves were attempting to launch an attack. Only GLM 5.2—which Hugging Face could run on its own servers—responded to the company's inquiries and provided the necessary defensive recommendations.

Hugging Face is one of the 37 companies involved in creating the alliance. Other members include Adobe, CrowdStrike, NVIDIA, Palo Alto Networks, and Siemens.

Similar to the open letter, the alliance acknowledges thatopen-source AI modelscarry real risks, including a lack of safeguards against cyberattack assistance. But the alliance states: "The right response is not to deny defenders access to capable open systems, but to combine openness with strong safeguards, clear rules against malicious misuse, rigorous evaluation, and rapid fixes."

The alliance says cyber defenders need both open-source and closed-source AI models to protect critical infrastructure and other vital computer systems.

The members wrote: "Comprehensive restrictions on open frontier AI systems would weaken defense capabilities and could concentrate power, dependency, and vulnerability in a few closed-source providers."