Anthropic CEO says spies are after $100M AI secrets in a ‘few lines of code’

Anthropic chief executive officer Dario Ambidei is concerned that spying from China is probably expensive from the top AI companies in the United States at the “Algorithmic Secrets” – and he wants the US government to take action.

Speech in a council related to foreign relations Event On Monday, Amedei said that China is known for its “large -scale industrial spy” and AI companies, including anthropologists, are almost targeting.

“There are many of these algorithmic privacy, Million is 100 million privacy that are a few lines of the code,” he said. “And, you know, I’m sure people are trying to steal them and they can succeed.”

The US government has added “very important” to protect against this risk, without mentioning exactly what kind of help will be needed.

Anthropic refused to comment on the techCranch, especially in comments, but mentioned Ethnographic recommendation Earlier this month, the Science and Technology Policy (OSTP) office of the House (OSTP).

In the case of submission, anthropological argument has shown that the Federal Government’s AI industry should be part of the US intelligence agency and their allies to be a partner for protection beef in Frontier AI labs.

These comments are in line with Amody’s more critical position on Chinese AI development. Amuse Call Strong US export controls in AI chips in China DIPSEC scored “worst” while saying The ethnographic ran in a critical biopon data protection test.

Amodi’s concern, as he put in his essay “Loving graceAnd somewhere else, the AI ​​for autocratic and military purposes focuses on China.

This type of position has taken Criticism Some of the AI ​​community who argue in the United States and China should provide less cooperation in AI, to avoid competition for a weapon, which can create a strong system for both countries so that people cannot control it.