Anthropic and the Pentagon: Breakthrough in AI and Security

Conflict Between Anthropic and the Department of Defense

While global attention was focused on Iran, an intense debate was unfolding in the tech sector in the United States. The federal government decided to end its collaboration with Anthropic, an artificial intelligence company, labeling it as a risk to the supply chain. This decision marks a significant escalation in the conflict between the government and Anthropic regarding the use of its technology by the Pentagon.

Meanwhile, OpenAI, a competitor of Anthropic, announced that it had reached an agreement with the Department of Defense. This agreement allows OpenAI to use its models in classified contexts, a privilege previously reserved for Anthropic. These rapid developments between the Pentagon and these two Silicon Valley companies could influence the future of advanced AI tool usage by the federal government, particularly within the Pentagon.

Anthropic's Position on AI Usage

Dario Amadei, a representative of Anthropic, expressed the company's stance in a recent statement. Anthropic believes that in certain cases, AI can undermine democratic values rather than uphold them. Two specific uses of AI, according to Anthropic, should not be included in contracts with the Department of Defense.

Domestic Mass Surveillance

Anthropic supports the use of AI for intelligence and counterintelligence missions abroad. However, the company firmly opposes the use of these systems for mass surveillance within the country, deeming it incompatible with democratic values. AI-powered mass surveillance poses serious risks to fundamental freedoms. Currently, the government can legally purchase detailed data on the movements and online activities of Americans without a warrant, a controversial practice that has raised privacy concerns and bipartisan opposition in Congress. Powerful AI can compile these scattered, individually harmless data points into a comprehensive picture of anyone's life—automatically and at scale.

Fully Autonomous Weapons

Regarding autonomous weapons, Anthropic acknowledges the importance of partially autonomous weapons for the defense of democracy, such as those used in Ukraine. However, the company believes that current AI systems are not reliable enough to power fully autonomous weapons, which remove humans from the target selection and engagement process. Even fully autonomous weapons could prove critical for our national defense. Anthropic refuses to provide products that could endanger American soldiers and civilians. The company has proposed collaborating with the Department of Defense to improve the reliability of these systems, but this offer has not been accepted. Without adequate oversight, fully autonomous weapons cannot be used safely.