Technology, cybersecurity and artificial intelligence companies have launched a coordinated campaign urging the U.S. government to support open-weight AI systems.
The push began on Friday when 76 companies signed an open letter promoting these models. Additionally, 37 companies launched the Open Secure AI Alliance on Monday to advance many of the same goals.
The companies argue the systems improve security, encourage competition and strengthen the country’s position against China’s rapidly advancing AI industry.
Furthermore, the efforts come as Chinese developers continue narrowing the performance gap with leading American AI systems. Recent advances have intensified debate in Washington and Silicon Valley over whether the United States should prioritize open or closed AI development.
The companies argue that open-weight models offer broader economic and security benefits than closed alternatives. Open-weight systems make their trained model parameters available, allowing researchers and businesses to study, modify and build upon existing technology.
According to the letter, U.S. leadership will depend on creating an open AI ecosystem rather than relying on a handful of advanced models. The signatories argued that widespread access would encourage innovation across industries while creating new opportunities for businesses and researchers.
Additionally, the companies said open-weight models lower barriers for startups and smaller developers. They argued that customers also gain greater flexibility because they can switch providers more easily instead of becoming dependent on one company’s technology.
The letter acknowledged that open-weight AI carries unique risks. Once released, anyone can modify the models, creating versions that could potentially support malicious activity or bypass original safeguards.
Read more: South Korea’s AI bubble bursts, stock market plunges with US$2T lost
Read more: Chipmaker CXMT becomes Mainland China’s top stock in explosive IPO
Open-weight supporters oppose AI concentration
However, the companies argued that openness also strengthens cybersecurity. They said researchers, developers and security teams can examine systems more closely, identify weaknesses faster and develop fixes before vulnerabilities become widespread.
The signatories also warned that closed AI systems introduce different security concerns. They argued that hidden vulnerabilities may remain undiscovered longer because outside experts cannot inspect the underlying technology.
Furthermore, the group cautioned against concentrating advanced AI capabilities among only a few providers. It argued that broader participation creates a more resilient technology ecosystem while reducing dependence on individual companies.
The open letter attracted support from several influential technology firms, including Cisco Systems (NASDAQ: CSCO), Alphabet Inc (NASDAQ: GOOGL), through its Google division, <strong>Microsoft Corp. (NASDAQ: MSFT), venture capital firm Andreessen Horowitz and OpenAI.
Meanwhile, many of the same organizations joined the newly formed Open Secure AI Alliance. The alliance plans to promote secure development practices while encouraging policymakers to recognize open-weight AI as a key component of America’s long-term technology strategy amid increasing competition from China.
The debate centres on whether AI systems should remain proprietary or be made more accessible after release, an important distinction that shapes the industry’s competing visions.
Artificial intelligence models generally fall into two categories: open-weight and closed. The difference comes down to how much access developers receive after a model is released.
Read more: Artificial Intelligence and the invisible revolution: A Mugglehead roundup
Read more: Tempus AI buys Personalis to strengthen AI-driven cancer monitoring platform
Open-weight models encourage competition
A closed model keeps its underlying technology private. Users can interact with the AI through a website or application programming interface, but they cannot see or modify the model’s trained parameters, known as weights. The company that built the model controls updates, pricing, access and security. Popular examples include OpenAI’s GPT models and Anthropic’s Claude.
An open-weight model makes those trained weights available for others to download and use. Developers can run the model on their own computers or cloud infrastructure, fine-tune it for specialized tasks and build new applications without relying on a single provider. However, open-weight models do not necessarily reveal everything used to create them, such as the training data or development process.
Supporters argue that open-weight models encourage competition, lower costs and improve security by allowing more researchers to inspect systems for weaknesses. Additionally, they say organizations gain greater flexibility because they are not locked into one vendor.
However, critics contend that open-weight models can also create new risks. Because anyone can modify and redistribute them, malicious actors may remove safety features or adapt them for cyberattacks, fraud or other harmful activities. As a result, the debate centres on whether broader access improves security through transparency or increases danger by making powerful AI tools easier to misuse.
.
