Dario Amodei on AI safety
From Code with Claude 2026: A conversation with Dario Amodei & Daniela Amodei · · Techusiness
“I think when I think about what Anthropic is trying to do, there's these sort of two pillars. The first is around how do we develop this transformative technology in a way that is good for everybody. And I think this goal of Claude is this incredible tool. It has the power to really transform what people build and how they create and the ambition level of what they can develop themselves. And I think there comes a huge amount of opportunity there. And there's also some risk, right? There's some risk to just labor disruption. There's risks to ensuring that the technology is developed safely that it's good for people.”
On , Dario Amodei, CEO and Co-Founder at Anthropic, spoke about AI safety during Code with Claude 2026: A conversation with Dario Amodei & Daniela Amodei on Techusiness.
About Dario Amodei
Dario Amodei, CEO and co-founder of Anthropic, has been publicly discussing the rapid growth of his company and the broader economic implications of artificial intelligence. In recent appearances, he stated that Anthropic's revenue has been increasing tenfold per year, approaching a $7 billion run rate, and that the company's growth has accelerated beyond the exponential trend. He also noted that Anthropic's valuation has been reported at $350 billion. Amodei described the company's trajectory as a "roller coaster" and said that the experience of seeing the scaling laws he helped predict play out in practice has been "surreal." Amodei has addressed the potential for AI-driven job displacement and economic disruption. He said that AI could lead to a combination of high GDP growth and high unemployment, a scenario he described as unprecedented. He predicted that within six months, 90% of code would be written by AI models, a claim he said has already proven true within Anthropic and partner companies. However, he argued that under comparative advantage, this would not reduce the need for software engineers in the short term, as they could focus on editing and supervision. He also warned that individual SaaS companies could go bankrupt if they fail to adapt, and that Chinese AI models would catch up within six to twelve months. On regulation, Amodei expressed skepticism about an FDA-style approval process for AI models, instead favoring a model of ongoing monitoring similar to the National Transportation Safety Board. He also voiced concern that AI could be "uniquely well suited to autocracy" by enabling surveillance, propaganda, and suppression of dissent.