Trump Invokes Defense Production Act to Force Unrestricted Access to Anthropic

The Trump administration’s decision to target Anthropic represents an unprecedented escalation in the federal government’s relationship with the artificial intelligence sector. By invoking the Cold War-era Defense Production Act (DPA), the administration is attempting to bridge a paradoxical gap: labeling Anthropic’s technology as a dangerous supply chain risk that other companies must avoid, while simultaneously declaring it so essential to national security that the government must have it without any safety restrictions or 'caveats.' This move follows the recent release of Claude Code, a tool that demonstrated such significant capabilities in software engineering that it was credited with wiping over $1 trillion in market value from the global software sector in a single month.

The legal strategy employed by the administration is two-pronged and highly aggressive. First, by designating Anthropic as a supply chain risk, the government is effectively blacklisting the company from the broader defense and federal contracting ecosystem. This puts immense pressure on Anthropic’s partners and enterprise clients, who may be forced to choose between using Claude and maintaining their own lucrative government contracts. Second, the threat to use the DPA to compel the delivery of 'caveat-free' versions of Claude suggests a desire to strip away the safety filters and ethical guardrails that Anthropic has spent years developing. This is a direct challenge to the 'Constitutional AI' framework that defines Anthropic’s brand and product philosophy.

“

This puts immense pressure on Anthropic’s partners and enterprise clients, who may be forced to choose between using Claude and maintaining their own lucrative government contracts.

From a RegTech and legal perspective, this development tests the limits of executive authority over intangible intellectual property. While the DPA has historically been used to prioritize the manufacturing of physical goods like steel or medical supplies during crises, its application to software code and algorithmic weights is largely uncharted territory. If the administration succeeds, it could set a precedent where the government can effectively nationalize or force the modification of any software deemed 'strategically vital.' This would fundamentally alter the risk profile for AI startups, who must now weigh the benefits of creating powerful tools against the risk of those tools being seized or regulated out of their control by the state.