Congress Considers New Rules on Military AI Use
In a significant move, Congress is actively debating new legislation that would restrict how the Pentagon can use artificial intelligence. The discussion around military AI regulation has gained urgency as lawmakers and technology companies clash over ethical and security boundaries for deploying these advanced technologies in defense operations.
Bipartisan Efforts to Shape Military AI Policy
Senator Kirsten Gillibrand, a Democrat from New York, has introduced a bill aimed at strengthening oversight of the Pentagon’s use of artificial intelligence. She plans to submit the proposals as amendments to the National Defense Authorization Act, the key annual law that sets the Pentagon’s budget and policies. The House Armed Services Committee is currently reviewing its version of the bill, while the Senate will begin its markup next week.
There is a growing push from Democrats for more robust military AI regulation, and they are seeking bipartisan support. Notably, Vice President JD Vance has signaled support for the core principles of these regulatory efforts, lending hope for a measure of consensus across party lines.
Key Provisions in the Proposed Legislation
Another bill, authored by Senator Elissa Slotkin of Michigan, is expected to become part of the Senate’s defense authorization package. Slotkin’s legislation would codify certain protections sought by artificial intelligence companies as they negotiate with the Pentagon. These include outright bans on using AI for domestic surveillance, controlling autonomous drones, or launching nuclear weapons.
While technology giants such as OpenAI and Google have accepted the Defense Department’s standard of “any lawful use,” the company Anthropic has resisted, arguing that current agreements do not provide enough safeguards against misuse. Anthropic’s disagreement with the Pentagon led to the company being labeled a “supply-chain risk to national security,” potentially excluding it from federal defense contracts. Anthropic has since filed a lawsuit to challenge this designation.
Exceptions and Oversight Mechanisms
Senator Gillibrand’s proposal does permit certain exceptions, allowing the use of AI in areas like cybersecurity, missile defense, base and ship protection, and semi-autonomous weapons. However, the bill requires the Defense Department to seek congressional approval for most other uses of autonomous weapons in combat scenarios, reinforcing the need for military AI regulation to ensure ethical safeguards.
“Lethal decisions require a conscience, not just an algorithm,” Senator Gillibrand stated. She emphasized that while AI can be a crucial decision-making tool, it lacks the human qualities necessary for life-and-death judgments, such as empathy and moral consideration.
Human Control Over Lethal Force
The issue of maintaining human involvement in military strike decisions has attracted bipartisan agreement. Vice President Vance, during a commencement speech at the Air Force Academy, highlighted the moral imperative that decisions involving lethal force should remain under human control, referencing a recent papal encyclical on the topic.
The Pentagon’s current policy on autonomous weapons, often described as requiring a “human in the loop,” mandates that a person must participate in decisions to carry out lethal strikes. However, the policy is nuanced and does not impose an outright ban on autonomous weapons. Instead, it requires senior-level review before any such system is developed or deployed. Congress has also mandated reporting on any use of lethal autonomous weapons through 2029.
Updating Policy for Rapid Technological Change
In a recent Senate hearing, Pentagon technology chief Emil Michael acknowledged that advances in adversaries’ capabilities and lessons learned from recent conflicts necessitate updating existing policy. Senator Jodi Ernst noted that the Pentagon is proposing $55 billion in spending on its Defense Autonomous Warfare Group, which develops both autonomous and semi-autonomous weapons. She highlighted that current guidelines may not adequately address innovations such as AI-driven targeting systems.
Protecting Americans’ Privacy and Data
Beyond autonomous weapons, the proposed legislation includes strict provisions to protect Americans’ privacy. AI labs would be required to report incidents where their models are stolen, their supply chains are compromised, or their data is corrupted. The bill imposes tough restrictions on using AI for domestic surveillance, explicitly banning the use of commercial or hacked data to analyze information about American citizens. It also bars the Pentagon and military intelligence agencies from creating AI models that assign risk scores to individuals or make predictive threat assessments.
Setting a Global Example
While Chinese companies are known to be researching predictive surveillance technologies, Senator Gillibrand’s measure aims to ensure that the U.S. military does not engage in similar activities. She views the recent dispute between Anthropic and the Pentagon as a warning sign that domestic use of AI in surveillance is a real concern.
“We made sure that under no circumstances should AI be able to surveil American citizens under the Department of Defense,” Gillibrand explained, underscoring the importance of comprehensive military AI regulation for national security and civil liberties.
Conclusion
The debate over military AI regulation reflects both the promise and peril of artificial intelligence in defense. As Congress moves forward, the goal remains clear: to harness the benefits of AI while enshrining safeguards that protect human rights, ethical standards, and national security.
This article is inspired by content from Original Source. It has been rephrased for originality. Images are credited to the original source.