WASHINGTON, D.C. — A United States appeals court has upheld the Pentagon’s decision to ban AI startup Anthropic from its supply chain after the firm refused to remove safeguards on its artificial intelligence models, in a ruling that underscores the growing tension between national security imperatives and AI safety concerns.
The ruling, issued on Thursday by the US Court of Appeals for the Federal Circuit, affirmed a lower court decision that had dismissed Anthropic’s challenge to the Pentagon’s ban. The court found that the Department of Defense had acted within its authority when it excluded the company from its procurement processes, citing national security concerns.
The dispute arose after the Pentagon demanded that Anthropic remove certain safety guardrails from its AI models, which the military said limited their operational effectiveness. Anthropic reportedly refused, arguing that the safeguards were essential to preventing misuse and ensuring that its AI systems remained safe and aligned with human values.
The Pentagon subsequently barred Anthropic from its supply chain, a move that effectively excluded the company from lucrative defence contracts and partnerships. Anthropic challenged the ban in court, arguing that it was arbitrary, capricious, and violated its constitutional rights.
But the appeals court disagreed, ruling that the Pentagon had a legitimate interest in ensuring that the AI systems it procures meet its operational requirements and do not contain features that could compromise national security.
“The Department of Defense has broad discretion to determine the terms and conditions under which it procures goods and services, particularly when national security is at stake. The court finds no basis to overturn the Department’s decision,” the ruling stated.
The court also rejected Anthropic’s argument that the ban violated its First Amendment rights, noting that the company’s refusal to remove the safeguards was a business decision, not an exercise of protected speech.
The ruling is a significant victory for the Pentagon, which has been pushing for greater access to cutting-edge AI technologies while insisting that its contractors comply with its operational and security requirements. It is also a setback for Anthropic, which has positioned itself as a leader in AI safety and has been vocal about the need for responsible AI development.
In a statement, Anthropic expressed disappointment with the ruling but said it remained committed to its mission of developing safe and beneficial AI. “We respect the court’s decision, but we continue to believe that safety guardrails are essential to the responsible development and deployment of AI. We will continue to advocate for policies that prioritise safety and accountability in AI,” the company said.
The Pentagon welcomed the ruling, saying it affirmed the Department’s authority to set the terms of its procurement processes. “The Department of Defense is committed to ensuring that the AI systems it acquires meet the highest standards of safety, security, and operational effectiveness. Today’s ruling affirms our authority to do so,” a Pentagon spokesperson said.
The case has drawn widespread attention in the tech and defence communities, highlighting the complex ethical and legal questions raised by the military’s use of AI. It also underscores the growing importance of AI safety as a factor in government procurement decisions, with companies increasingly being asked to balance innovation with responsibility.
As the use of AI in military applications continues to expand, the tension between national security and AI safety is likely to intensify. For now, the ruling is a clear signal that the Pentagon will prioritise its operational requirements over the concerns of AI companies that refuse to comply with its demands. For Anthropic, the decision is a setback, but the company has vowed to continue its mission of developing AI that is safe, trustworthy, and beneficial to humanity.
