White House officials reportedly frustrated by Anthropic’s law enforcement AI limits

Anthropic's advanced AI models have the potential to assist intelligence agencies in analyzing sensitive documents, yet the company has firmly opted against allowing their application in domestic surveillance. This decision has reportedly sparked frustration within the Trump administration. According to a report by Semafor, officials within the White House are expressing their displeasure over Anthropic's strict limitations on the use of its Claude AI models for law enforcement purposes. Senior officials speaking anonymously revealed that federal contractors collaborating with agencies such as the FBI and Secret Service have encountered significant obstacles when trying to deploy Claude for surveillance operations. The root of the tension lies in Anthropic's policies, which explicitly ban the use of its technology for domestic monitoring. Concerns have been raised by the officials about the possibility that these policies are applied inconsistently and are influenced by political factors, with vague language allowing for broad interpretations of the rules. These restrictions significantly impact private contractors who are seeking AI solutions for law enforcement duties. In certain instances, Anthropic's Claude models are reportedly the only AI systems authorized for use in highly classified security contexts, thanks to Amazon Web Services' GovCloud. Additionally, Anthropic has established a specialized service for national security clients and has struck a deal with the federal government to provide its services at a nominal fee of $1. While the company is also engaged with the Department of Defense, it maintains a stance against allowing its models to be used for any weapons development. This comes in the wake of OpenAI unveiling a competing initiative that provides access to ChatGPT Enterprise for over 2 million federal employees for just $1 per agency for one year, shortly after a General Services Administration agreement enabled OpenAI, Google, and Anthropic to supply their tools to federal workers.