Khaberni - Despite more than a month passing since the arrest of Venezuelan President Nicolas Maduro, some details are still gradually emerging about that «complex and dangerous» operation.
According to sources familiar with the «Axios» website, the U.S. military used the artificial intelligence model Claude from Anthropic company during the operation that targeted the arrest of Venezuelan President Nicolas Maduro.
The American website says that the role played by «Claude» in this «complex and dangerous» operation, which was not previously announced, highlights the tensions facing major AI labs when they engage in dealings with military institutions, as these companies strive to maintain some restrictions on how their tools are used.
Intelligent models like «Claude» are capable of processing data quickly in real-time, a capability valued by the Pentagon given the chaotic environments in which military operations take place.
«Axios» was unable to verify the exact role played by «Claude» during the operation, but sources said that the model was used not only in the preparations but also during the execution of the operation itself. No American casualties were recorded during the attack, while Cuba and Venezuela announced dozens of their soldiers had fallen.
The U.S. Department of Defense wishes that major AI companies would allow their models to be used in any scenario as long as it complies with the law, while «Anthropic», which markets itself as a leader in safe usage, seeks to negotiate with the Pentagon the terms of use, especially emphasizing not using its technology in mass surveillance of Americans or to operate autonomously functioning weapons.
The source stated that the company is confident that the military complied with the current use policies in all instances.
In turn, a spokesperson for «Anthropic» said in statements to the «Axios» website: «We cannot comment on whether «Claude», or any other artificial intelligence model, was used in any particular operation, whether secret or otherwise».
He added: «Any usage of «Claude» — whether in the private sector or across governments — must comply with our own use policies, which govern how the model is deployed. We work closely with our partners to ensure compliance».
U.S. Defense Secretary Pete Hegseth has pushed for the rapid integration of artificial intelligence into various aspects of military work to gain an advantage over China, according to official statements, but the Pentagon did not respond to a request for comment.
«Anthropic» is one of several major companies manufacturing AI models that cooperate with the U.S. Department of Defense in multiple ways, alongside OpenAI, Google, and xAI, all of which have agreements that allow military users to access their models with some eased use guarantees.
However, the most sensitive work—from weapons testing to communications during operations—is done on classified systems, and currently, only Anthropic's system is available on those classified platforms.
It is worth noting that Anthropic also has a partnership with Palantir, which has extensive contracts with the Pentagon, allowing the use of «Claude» in its security products, but it is unclear whether the use of «Claude» in the operation is linked to this partnership.
Discussions are ongoing between the Pentagon and OpenAI, Google, and xAI to allow the use of their tools on classified systems, as well as discussions between Anthropic and the Pentagon about potentially easing the restrictions on the use of «Claude» in the future.
