Anthropic and Palantir bring Claude to U.S. defense agencies

Anthropic, the AI lab known for championing safety and responsible development, is officially heading inside the Pentagon's classified walls.

The startup announced a partnership with big data giant Palantir and Amazon Web Services (AWS) to bring its Claude 3 and 3.5 model families to U.S. defense and intelligence agencies. The models were integrated into Palantir’s Artificial Intelligence Platform (AIP) on AWS earlier this month.

Entering Secret-level clearance

Deploying frontier models in government operations usually hits a brick wall: security clearance. Intelligence analysts cannot just feed top-priority data into public endpoints.

This partnership clears that hurdle through Palantir’s Impact Level 6 (IL6) accredited environment, powered by AWS infrastructure and Amazon SageMaker. IL6 is reserved for systems vital to national security, handling data classified up to Secret—just one step below Top Secret. Palantir and AWS are among only a handful of vendors holding this strict accreditation.

Defense and intelligence teams will use Claude to analyze massive troves of sensitive data, spot hidden patterns, review documents, and support rapid decisions. Anthropic's terms explicitly permit use cases such as authorized foreign intelligence analysis, detecting covert sabotage campaigns, and providing advance warnings of military activities.

To show what the stack can do, Palantir CTO Shyam Sankar cited a commercial client: a major U.S. insurer deployed 78 AIP- and Claude-powered agents to automate underwriting, slashing a two-week workflow down to three hours. Palantir is pitching that exact operational speedup to military operators.

Big AI turns toward defense

Anthropic is hardly alone in chasing public sector revenue. Meta recently cleared defense contractors to use its open-source Llama models, while OpenAI is reportedly pursuing deals with U.S. defense entities.

The expansion also aligns with Anthropic’s business backing. Amazon is already Anthropic’s largest investor, and Anthropic is reportedly in talks for a new funding round valuing the company at $40 billion.

Why it matters

Anthropic played this move smartly. Getting deep into classified government contracts directly is an administrative nightmare that takes startups years of compliance audits. By piggybacking on Palantir’s existing IL6 clearance and AWS's government-grade cloud, Anthropic bypassed the red tape and plugged straight into active defense budgets.

For the AI landscape, it signals that the enterprise battleground has officially expanded into national security infrastructure. With Meta offering open-source Llama to defense contractors and Anthropic teaming with Palantir, competitors like OpenAI will face immense pressure to formalize their own military relationships.

However, this shift carries real reputational risk. Anthropic built its brand on Constitutional AI and strict harm reduction. While the company insists human operators maintain full decision-making authority, navigating the blurry line between defensive intelligence analysis and active military operations will test the credibility of its responsible AI ethos.