Cohere CEO says foreign AI access poses national security risk

The CEO of Canadian AI company Cohere stated that governments face a national security risk when they rent AI capabilities from foreign providers rather than control the technology themselves. "This sentiment of renting AI from someone rather than owning it is a national security risk," the CEO said.

" The CEO said he had planned for the possibility that foreign governments could cut off access to frontier AI models. Earlier this month, the U.S. government moved to restrict foreign access to Anthropic’s Mythos, which the CEO described as an example of the concern.

Infrastructure control The CEO said centralized dependence on a single entity creates a structural risk because access can be revoked and services shut down. He argued that democracies should build sovereign systems they control end-to-end, including data centers, chips, and models.

Cohere sells its models so governments and companies can run them on their own infrastructure, including systems that are physically disconnected from the internet and to which the company has no remote access.

Limits of sovereignty The CEO said the term “sovereign AI” lacks a settled definition, with some governments claiming progress without committing to full independence. He said domestically controlled infrastructure is the first requirement for sovereignty.

Most countries cannot own every layer of the AI stack due to high costs for compute, capital, and energy, the CEO said. He argued that democracies should instead pool resources around a small number of champions and share capacity through trusted partnerships.

Cohere announced a deal in April to acquire Germany’s Aleph Alpha, aiming to create a transatlantic company. The combined firm would retain the Cohere name and be valued at around $20 billion, according to the Financial Times. At last week’s G7 discussions, the CEO said leaders and AI executives agreed on the need for democratic countries to coordinate on standards and regulation.