Days after Chinese lab Moonshot’s Kimi K3 model made headlines due to its specifications and open-weight labelling, AI giants in the U.S.—excluding Anthropic—released an open letter that stressed the importance of American open-weight models.
The letter, dated July 24 and co-signed by over 60 companies, analysed both the advantages and disadvantages of open-weight models when compared to their closed counterparts. It ultimately concluded that open-weight models would make advanced AI more accessible, adaptable, and available to American users, as opposed to locking users to specific providers.
The letter came soon after the launch of Kimi K3 ignited user enthusiasm over what the Chinese company claimed was the world’s first open model in the 3-trillion-parameter class (2.8 trillion parameters, to be specific).
Kimi K3, though tagged as an open-weight model (which is different from an open-source model), released its full model weights on July 27.
While Anthropic has accused Moonshot of distilling its models to improve its own offerings, the release of Kimi K3 was met with admiration, sparking debates over Chinese open-weight AI models potentially disrupting the dominance of closed AI models coming from firms such as OpenAI and Anthropic.
From a cybersecurity perspective, the letter cited attackers who used advanced AI and noted that open-weight models supported cyber defence use cases while also increasing transparency and vulnerability detection.
“Our AI leadership will be judged not by one frontier AI model, but by whether the United States builds a strong, open ecosystem that diffuses into every sector. This is essential for creating opportunities for innovation and prosperity across the country,” stated the letter.
The letter called upon policy-makers to take action by expanding access to compute for startups and researchers, investing in shared training assets, and avoiding “premature restrictions” on open models in order to prevent the stifling of competition or driving innovation overseas.
Some high-profile tech companies that co-signed the letter included Microsoft, Nvidia, OpenAI, SpaceX, Google, IBM, Palantir, and Meta.
One noticeable absence was Google and Amazon-backed Anthropic. Amazon was also not listed as a signatory.
“AI safety may depend on giving more people the ability to test and strengthen the models on which society relies. It allows for rigorous benchmarking and evaluation, red teaming, and protections tied to real and demonstrated harms rather than assuming that closed systems are safer by default,” stated the letter, concluding that the United States should lead in building an AI age of “prosperity”.
On July 27, Anthropic CEO Dario Amodei published a blog post clarifying the company’s stance on open-weight models, highlighting that Anthropic had never advocated for a ban on open-weight models.
However, he detailed the risks of AI models being developed by hostile states and used for cyber-attacks or biological attacks (such as by weaponising pandemic-level viruses).
“Open-weights models—it does not matter whether they come from China or anywhere else—do potentially present a higher risk than closed models, because it is very difficult to apply guardrails to them or monitor their usage, and once weights are released they cannot be withdrawn,” the Anthropic CEO noted.
Mr. Amodei said instead that the U.S. should not sell powerful chips or chipmaking equipment to China, that the U.S. should address industrial-scale distillation, and that “sufficiently capable models” should go through mandatory safety testing.
While agreeing with most of the joint letter’s contents, he raised the question of whether open-weight models would truly be more beneficial to defenders than attackers. The initial letter had also acknowledged the risks of weights beyond the original developer’s control.
“To summarize my and Anthropic’s position, we have not and are not advocating for a ban on open-weights models as a category. We should instead focus on keeping powerful chips out of authoritarian hands, stopping industrial-scale distillation, and requiring safety testing of all sufficiently capable models, open and closed,” said Mr. Amodei.
Published – July 28, 2026 11:00 am IST
