Anthropic CEO Dario Amodei denies call to ban open-weight models
Make us a preferred source
Make us a preferred source on Google
The news: Anthropic CEO Dario Amodei has stated that the company “has never advocated for a ban on open-weights models” after it refused to sign an open letter backing the technology as the US considers restricting Chinese models.
The context: The open letter, released over the weekend, outlined the benefits of open weight models, which publicly disclose key development parameters, and cautioned against prohibition. It was signed by dozens of companies including Nvidia, OpenAI, SpaceX, Microsoft and Google.
Following the release of powerful new Chinese models rivalling the latest US offerings, the US government has been considering restricting access to the Chinese models.
While Anthropic has not signed the letter, Amodei said in a blog post released on Tuesday morning (AEST) that “open-weight models that don’t have dangerous capabilities are a public good: they don’t cost anything besides the compute needed to run them, and they provide value to business, developers, and researchers”.
Amodei said he agreed with much of the letter but not with its “assertions that open-weight models necessarily make it easier to develop safeguards or that broad access to capabilities necessarily helps defenders more than attacks”.
Rather than a ban on open weight models, Amodei said that “all sufficiently capable models, open and closed, should go through mandatory safety testing”, the US government should crackdown on distillation activity in which frontier models are used to train new models, and the chips and chipmaking equipment should be kept “out of authoritarian hands”.
Amazon Web Services CEO Matt Garman announced in a social media post on Tuesday morning that it had also signed on.
“While there are elements of the Open Weights and American AI Leadership letter that need further definition over time, we’re supportive of the macro direction and have signed it,” Garman said.
