A Clash Over AI Ethics: Anthropic Stands Firm Against US Government Demands
A significant dispute has erupted between the US Department of War and Anthropic, a leading American artificial intelligence company. The conflict centers on the company’s refusal to comply with two specific demands regarding the use of its AI technology. Anthropic’s CEO, Dario Amodei, has publicly rejected what he describes as legally inconsistent pressure from the government, asserting that threats will not force the company to alter its principles.
The tension escalated when Pete Hegseth, the head of the Department of War, directed that Anthropic be classified as a national security supply-chain risk. This label is typically reserved for foreign adversaries, not domestic companies. Anthropic has pushed back, arguing that such a designation would be historically unprecedented and legally flawed.
The Core of the Disagreement
Anthropic states that it fully supports all lawful uses of AI for US national security. However, the company has drawn a firm line on two exceptions: mass domestic surveillance of American citizens and the development of fully autonomous weapons systems that remove humans from the decision-making loop. The company insists that these two restrictions have never hindered any government mission or military operation to date.
According to Amodei, the Department of War wants Anthropic to provide unconditional consent for any statutory use of its technology. The company has refused, maintaining that these two ethical boundaries are non-negotiable. The standoff has now become a public test of how far the government can push private AI developers to comply with its security agenda.
The Supply-Chain Risk Controversy
Pete Hegseth has instructed his department to formally declare Anthropic a supply-chain risk to national security. This move, if enacted, would mark the first time such a label has been applied to a US-based company. Anthropic argues that under Title 10, Section 3252 of the US Code, this designation can only apply to specific defense contracts, not the entire private sector. Furthermore, the company contends that the Secretary of War lacks the legal authority to prohibit private contractors from doing business with Anthropic.
The company has called the threat legally inconsistent and warned that it could set a dangerous precedent for other American technology firms. By labeling a domestic AI leader as a risk, the government risks undermining the very innovation ecosystem it claims to protect.
Pressure Under the Defense Production Act
Beyond the supply-chain label, the Department of War has also signaled that it may invoke the Defense Production Act to compel Anthropic to drop its safeguards. The company has pointed out the inherent contradiction in this approach: on one hand, the government calls Anthropic a security risk, and on the other, it deems the company’s technology essential for national security under the same legal framework.
Amodei has described this dual stance as internally inconsistent. He questioned how a company could be both a threat and a necessity at the same time. This paradox, he argues, reveals a deeper tension in the government’s strategy toward AI regulation and procurement.
CEO Dario Amodei’s Official Statement
In a detailed public statement, Amodei laid out his position clearly. He expressed a deep belief that AI should be used to defend the United States and other democracies, calling it a matter of existential importance. He noted that Anthropic’s AI model, Claude, is already deployed for mission-critical tasks such as intelligence analysis, cyber operations, and operational planning.
Amodei also revealed that Anthropic had walked away from millions of dollars in revenue associated with the Chinese Communist Party, solely to protect American technological interests. Despite this record of cooperation, he emphasized that two issues remain absolute red lines:
- Mass domestic surveillance: Collecting data on Americans’ activities and web browsing undermines democratic values. Powerful AI systems can piece together scattered data to create comprehensive profiles of individuals’ lives, posing a direct threat to fundamental freedoms.
- Fully autonomous weapons: Current AI systems are not reliable enough to be given the authority to attack targets without human oversight. Amodei stated that Anthropic will not produce any product that puts American warfighters at risk by removing them from the decision-making process.
A Question of Risk or Necessity
Anthropic has raised pointed questions about the government’s contradictory messaging. Amodei highlighted two specific inconsistencies:
- The controversial label: Being declared a supply-chain risk, a designation historically reserved for foreign enemies, is legally untenable and sets a dangerous precedent for American industry.
- Conflicting threats: The government simultaneously labels Anthropic a security risk while using the Defense Production Act to force the company to remove its safeguards. If the company is truly a threat, Amodei asked, why is its technology considered so essential?
These questions underscore a broader debate about the balance between national security interests and ethical AI development. Anthropic appears willing to cooperate with the government, but not at the expense of its core principles.
Pete Hegseth’s Public Criticism
Before the official designation was announced, Pete Hegseth took to social media to criticize Anthropic. He described the company’s stance as a masterclass in arrogance and betrayal. Hegseth argued that the US military must have full and unrestricted access to AI models to defend the republic. According to him, Anthropic is prioritizing Silicon Valley ideology over American lives.
Hegseth has directed that no government contractor may do business with Anthropic going forward, though a six-month transition period has been granted to shift services. This directive, if enforced, could have significant financial implications for the company and its partners.
The unfolding dispute between Anthropic and the Department of War represents a pivotal moment in the relationship between the US government and the AI industry. It raises fundamental questions about ethics, national security, and the limits of corporate autonomy in an era of rapidly advancing technology. Both sides have drawn their lines, and the outcome of this confrontation could shape AI policy for years to come.
