Anthropic Joins Palantir and AWS to Provide Artificial Intelligence to Defense Sector Clients.

Anthropic announced a collaboration with Palantir and Amazon Web Services (AWS) to provide U.S. intelligence and defense agencies access to its family of AI models known as Claude. This initiative comes at a time when various AI providers are seeking to establish agreements with defense clients in the United States for both strategic and financial reasons. Recently, Meta also announced that it will make its Llama models available to its defense partners, while OpenAI is attempting to strengthen its relationship with the U.S. Department of Defense.

Kate Earle Jensen, head of sales at Anthropic, explained that the collaboration with Palantir and AWS aims to "operationalize the use of Claude" within Palantir's platform, using AWS's hosting infrastructure. Claude was integrated into Palantir's platform earlier this month, allowing its use in the accredited defense environment known as Palantir Impact Level 6 (IL6). This level from the Department of Defense is reserved for systems that handle data deemed critical to national security and that require "maximum protection" against unauthorized access and manipulation. IL6 systems may contain classified information, up to the level of "secret," which is one step below the highest level of confidentiality.

Jensen expressed Anthropic's pride in leading the provision of responsible AI solutions in classified environments in the U.S., which will enhance analytical capabilities and operational efficiency in essential government functions. The intention behind accessing Claude in Palantir through AWS is to provide defense and intelligence organizations with powerful AI tools that enable rapid processing and analysis of large volumes of complex data. This results in a significant improvement in intelligence analysis, optimizing decision-making and increasing operational efficiency across various agencies.

This summer, Anthropic introduced certain Claude models in AWS's GovCloud, a move that underscores its interest in expanding its client base in the public sector. GovCloud is an AWS service specifically designed for U.S. government workloads. Unlike OpenAI, Anthropic has positioned itself as a more security-conscious provider. However, its terms of service allow the use of its products for activities such as "authorized foreign intelligence analysis," "identifying hidden influence or sabotage campaigns," and "early warnings of potential military activities."

Anthropic also clarifies that it will adapt usage restrictions according to the mission and legal authorities of government entities, taking into account factors such as the agency's willingness to maintain a continuous dialogue. However, these terms do not apply to AI systems that pose a significant risk of catastrophic misuse, demonstrate low-level autonomous capabilities, or can be used for disinformation campaigns, weapons design or implementation, censorship, internal surveillance, and malicious cyber operations.

Government agencies are showing notable interest in AI. An analysis by the Brookings Institute conducted in March 2024 revealed a 1,200% increase in government contracts related to this technology. However, some branches, like the U.S. Army, have been slow to adopt these advances, maintaining skepticism regarding the associated return on investment. Anthropic, which has recently expanded its operations to Europe, is in discussions to secure a new funding round, with a valuation that could reach $40 billion, having raised around $7.6 billion so far, with Amazon being its largest investor.