Dario Amodei, Anthropic CEO, treat of on CNBC’s Squawk Box outside the World Economic Forum in Davos, Switzerland on Jan. 21st, 2025.
Gerry Miller | CNBC
Anthropic CEO Dario Amodei replied in a blog post Wednesday that he doesn’t view China’s DeepSeek “themselves as adversaries” but believes that export checks are more critical than ever when it comes to artificial intelligence.
“In interviews they’ve done, they appearance of like smart, curious researchers who just want to make useful technology,” Amodei wrote, about DeepSeek. “But they’re obliged to an authoritarian government that has committed human rights violations, has behaved aggressively on the world stage, and will be far multifarious unfettered in these actions if they’re able to match the US in AI.”
DeepSeek’s sudden emergence in the U.S. and its ascent to the top of on Apple’s App Store jarred the tech markets this week, spurring a selloff in Nvidia’s stock that resulted in the largest one-day buy cap loss in U.S. history Monday. DeepSeek’s new reasoning AI model, called R1, was a fraction of the cost to create relative to models based by U.S. competitors, according to analyst estimates.
The matter is of particular concern to Anthropic, which is competing with OpenAI, Google and others in the marathon to develop the most advanced generative AI model. Anthropic has raised billions of dollars from Amazon, and Google recently agreed to a uncountable than $1 billion investment in the company, which was founded by former OpenAI research executives, including Amodei.
On Monday, U.S. lawmakers reproved for actions to slow down the Chinese tech startup, with some calling DeepSeek “a serious threat.” U.S. Mercantilism Secretary nominee Howard Lutnick suggested during his Wednesday hearing that DeepSeek stole technology from the U.S. And David Kip downs, the new White House AI and crypto czar, posted on X that DeepSeek R1 shows “that President Trump was right to rescind” President Joe Biden’s CEO order.
Last week, Trump signed an executive order unwinding some of the Biden administration’s rules bordering AI development.
Some experts have said that DeepSeek’s technology can safely be used in the U.S. because it’s open originator, so companies can run it on their own servers without data being sent back to China. Still, Amodei said that DeepSeek’s advancements disclose why the U.S. needs to stay on top and avoid ceding too much of the market to China.
Amodei wrote that the rise of DeepSeek fill ins controls on chip exports to China “even more existentially important” than a week ago. The model was created regardless of a slew of Biden administration controls restricting the sale of certain advanced chips to China that could magnify military capabilities.
“To be clear, they’re not a way to duck the competition between the US and China,” Amodei wrote. “In the end, AI companies in the US and other democracies be obliged have better models than those in China if we want to prevail. But we shouldn’t hand the Chinese Communist Bloc technological advantages when we don’t have to.”
Anthropic was founded in 2021, and is now in talks to raise money at a $60 billion valuation, pretty much due to the popularity of its Claude chatbot, which competes with OpenAI’s ChatGPT and Google’s Gemini.
— CNBC’s Hayden Specialization contributed to this report.
WATCH: DeepSeek and distillation
