Anthropic Launches Claude Gov for U.S. Defense and Intelligence

On Thursday, Anthropic announced the launch of Claude Gov, a new artificial intelligence product specifically designed for U.S. defense and intelligence agencies. This innovative AI model features looser guardrails for government use and is trained to analyze classified information more effectively.

The company stated that the models being introduced are already in use by agencies at the highest levels of U.S. national security. Access to these models will be restricted to government agencies that handle classified information, although Anthropic did not disclose how long these models have been operational.

Claude Gov models are tailored to meet the unique needs of government operations, such as threat assessment and intelligence analysis, according to Anthropic’s blog post. While the company emphasized that these models underwent the same rigorous safety testing as all of its Claude models, they possess specific features designed for national security applications. For instance, they are programmed to “refuse less when engaging with classified information,” a contrast to the consumer-facing Claude models, which are trained to flag and avoid such content.

In addition, Claude Gov models exhibit a greater understanding of documents and context relevant to defense and intelligence, as well as enhanced proficiency in languages and dialects pertinent to national security.

The use of AI by government agencies has faced significant scrutiny due to potential harms and the impact on minorities and vulnerable communities. There have been numerous instances of wrongful arrests linked to police use of facial recognition technology multiple U.S. states, concerns regarding bias in predictive policing documented evidence, and discrimination arising from government algorithms assess welfare aid. Furthermore, there has been ongoing controversy surrounding major tech companies like Microsoft, Google, and Amazon providing AI products to military entities, particularly in Israel, which has sparked campaigns and public protests under the No Tech for Apartheid movement.

Anthropic’s usage policy explicitly states that users must not create or facilitate the exchange of illegal or highly regulated weapons or goods. This includes using Anthropic’s products or services to produce, modify, design, market, or distribute weapons, explosives, dangerous materials, or other systems intended to cause harm or loss of human life.

Approximately eleven months ago, the company said introduced a set of contractual exceptions to its usage policy, which are designed to enable beneficial uses by carefully selected government agencies. Certain restrictions remain in place, prohibiting disinformation campaigns, the design or use of weapons, the construction of censorship systems, and malicious cyber operations. However, Anthropic retains the ability to tailor use restrictions to the mission and legal authorities of a government entity, while striving to balance the enabling of beneficial uses with the mitigation of potential harms.

Claude Gov represents Anthropic’s response to ChatGPT Gov, a similar product launched by OpenAI for U.S. government agencies in January. This development is part of a broader trend among AI giants and startups seeking to strengthen their business relationships with government entities, particularly in an uncertain regulatory environment.

When OpenAI unveiled ChatGPT Gov, the company reported that over 90,000 employees from federal, state, and local governments had utilized its technology within the past year for various tasks, including document translation, summary generation, policy memo drafting, code writing, and application development. Anthropic, however, declined to disclose specific numbers or use cases but is involved in Palantir’s FedStart program, a software-as-a-service (SaaS) offering for companies aiming to deploy software for federal government use.

Scale AI, a prominent AI company that supplies training data to industry leaders such as OpenAI, Google, Microsoft, and Meta, signed a deal partnered with the Department of Defense in March to initiate a groundbreaking AI agent program for U.S. military planning. Since then, Scale AI has expanded its operations to global governments, recently securing a five-year contract with Qatar to provide automation tools for civil service, healthcare, transportation, and more.