Anthropic CEO says he does not support banning open-weight AI models

In short: Anthropic CEO Dario Amodei said his company has not advocated for a ban on open-weight AI models, even as he warns about security risks from more powerful AI.

What happened

Anthropic founder and CEO Dario Amodei published a blog post saying the company has “never advocated for a ban on open-weights models.” He wrote the post after industry chatter suggested Anthropic might support government efforts to restrict open-weight models, including ones from China.

Open-weight models are AI systems where the “weights” (the trained settings that act like the model’s memory and skills) are shared publicly. You can think of weights like the parts inside a device that make it work, not just the on and off button.

Amodei said open-weight models that do not have “dangerous capabilities” can be a public good, because many people can use them as long as they have enough computing power to run them. He also said he is not worried about businesses using open-weight models, including models from China.

His bigger concern is that authoritarian governments could build AI that is more powerful than what the United States has, and use it for military advantage or to control their own people. He also said he worries AI could enable biological attacks, and argued that open-weight models can be harder to put safety limits on or monitor. He noted that once open weights are released, they are difficult to take back.

Why it matters

This debate affects how easily companies, researchers, and hobbyists can access AI they can run themselves, instead of relying on a few large providers. It also shows a growing split between two worries, sharing AI widely for innovation, and limiting access to reduce misuse, especially in high-stakes areas like security and bio risk.

Source: TechCrunch AI