Senator Pushes for AI Guardrails in Military Operations Amidst Growing Concerns
A prominent US senator has put forward new legislation aimed at curbing the military’s use of artificial intelligence, specifically prohibiting its deployment for surveillance of American citizens or the authorisation of lethal strikes without direct human command. The proposed bill, dubbed the “AI Guardrails Act,” was introduced by Senator Elissa Slotkin of Michigan, a former CIA analyst.
Senator Slotkin’s initiative comes at a time of rapidly increasing AI integration in global conflicts and a perceived lack of comprehensive regulations surrounding this fast-evolving technology. She voiced her concerns during a recent Armed Services Committee hearing, stating, “If we were a healthier country politically right now, we would be putting up left and right limits around the use of AI. We have really no guidelines for you all, really no law. That’s not your fault. That’s on us, on a bipartisan basis.”
The proposed legislation would also explicitly ban AI-driven decisions to initiate a nuclear strike, a move that would significantly restrict the potential for an automated escalation of global conflict.
The Department of Defense has responded to the proposed bill, with officials asserting that their current practices already align with the spirit of the guardrails being sought. A statement released last month by Defense Department official Sean Parnell addressed the matter directly: “The Department of War has no interest in using AI to conduct mass surveillance of Americans (which is illegal) nor do we want to use AI to develop autonomous weapons that operate without human involvement.” This statement followed reports of stalled negotiations with AI firm Anthropic over ethical considerations surrounding autonomous weapons.
The U.S. military’s engagement with AI in recent operations, particularly in the context of the Iran conflict, has been extensive. A significant tool highlighted in this regard is Palantir’s Maven system. This system, when integrated with large language models like Anthropic’s Claude, plays a crucial role in analysing vast quantities of intelligence data and mapping information. It provides commanders with real-time battlefield insights and assists in the identification of potential targets.
Admiral Brad Cooper of Central Command recently elaborated on the benefits of these AI tools, stating, “These systems help us sift through vast amounts of data in seconds, so our leaders can cut through the noise and make smarter decisions faster than the enemy can react. Humans will always make final decisions on what to shoot and what not to shoot, and when to shoot. … But advanced AI tools can turn processes that used to take hours and sometimes even days into seconds.”
Silicon Valley’s Deepening Involvement in Defence AI
The influence of major technology companies from Silicon Valley in the defence sector is becoming increasingly pronounced. Beyond Palantir and Anthropic, other key players such as OpenAI, Google, Elon Musk’s xAI, and Anduril are either currently supplying or have agreements in place to provide the U.S. with a range of AI-driven defence systems.
Scrutiny Over Targeting and Ethical Dilemmas
The U.S. military’s targeting processes have faced intense scrutiny following an incident where a strike, believed to be American, tragically hit a girls’ primary school in Iran, resulting in the deaths of at least 175 people.
In the lead-up to the Iran conflict, the Pentagon engaged in protracted discussions with Anthropic regarding contract terms for the integration of its AI technologies into defence systems. Anthropic has claimed that the Defense Department sought unrestricted usage of their AI capabilities. The company, in turn, sought assurances that its products would not be employed for mass surveillance or as fully autonomous weapons. The Pentagon has maintained that it assured Anthropic that its models would only be utilised for lawful purposes.
However, these negotiations ultimately broke down. Anthropic CEO Dario Amodei stated last month that the company “cannot in good conscience” accept such terms. This decision led Defense Secretary Pete Hegseth to designate Anthropic as a supply chain risk, prompting the Trump administration to issue a directive for federal agencies to cease using Anthropic technology within six months. Anthropic has since initiated legal action to challenge this designation, arguing that the administration’s actions are politically motivated.
The debate surrounding the ethical implications of AI in warfare, particularly concerning autonomous weapons and the potential for civilian casualties, underscores the urgency of establishing clear guidelines and robust oversight mechanisms. Senator Slotkin’s proposed AI Guardrails Act represents a significant step towards addressing these critical concerns and ensuring that technological advancements in defence remain aligned with fundamental ethical principles and legal frameworks. The ongoing dialogue between technology providers, military leaders, and policymakers will be crucial in shaping the future of AI in national security.
