Anthropic, an AI company, announced on Thursday that it would not grant the US Defense Department unrestricted use of its technology, despite facing pressure from the Pentagon to do so. Chief executive Dario Amodei stated that they could not, in good conscience, comply with the request, emphasizing that using their technology for mass surveillance of US citizens and fully-autonomous weapons goes against their ethical standards.
The Pentagon had set a deadline for Anthropic to agree to unconditional military use of its technology by Friday, warning of potential enforcement under emergency federal powers if they did not comply. Amodei highlighted that while Anthropic’s models have been utilized by the Pentagon and intelligence agencies for defense purposes, they draw a line when it comes to applications involving mass domestic surveillance and autonomous weaponry.
Amodei stressed that deploying AI systems for mass surveillance within the country contradicts democratic values. He also noted that the reliability of leading AI systems is not yet at a level where they can be entrusted to power lethal weapons without human oversight, stating that they refuse to provide products that could endanger American military personnel and civilians.
Following discussions with Anthropic earlier in the week, the Pentagon issued a strong ultimatum, demanding compliance with unrestricted military use of the technology by a specific deadline or face potential enforcement under the Defense Production Act, a law granting the government broad powers to prioritize national security needs.
Moreover, the Pentagon threatened to designate Anthropic as a supply chain risk, a move typically reserved for companies from adversary nations, which could significantly impact Anthropic’s collaborations with the US government and its reputation. Despite the Pentagon’s assertions that its orders are lawful and within regulations, Anthropic remains firm in its stance on ethical considerations surrounding the use of its technology.
Additionally, it was confirmed that discussions regarding intercontinental ballistic missiles had occurred between Anthropic and the Pentagon, emphasizing the critical nature of the technologies at the core of the dispute. While Elon Musk’s Grok system received clearance for classified use, other companies like OpenAI and Google were reported to be close to similar approvals, placing competitive pressure on Anthropic to conform.
Last year, Anthropic, alongside other companies, secured a $200 million agreement to provide AI models for various military applications. Founded by former OpenAI employees with a focus on prioritizing safety in AI development, Anthropic’s philosophy has now brought it into conflict with the Pentagon and the White House, as they believe AI, in certain contexts, could undermine democratic values despite the military’s authority in decision-making.
Amodei affirmed that while the Department of War holds the responsibility for military decisions, they believe that AI, in specific instances, could pose a threat to democratic principles rather than safeguarding them.