AI Warfighting Is Dangerous, Lethal, Costly Without Protections

Killer robots threaten to become the immoral and destabilizing norm unless urgent action is taken to curtail the military use of artificial intelligence (AI), according to a Public Citizen report released today.

The report, “A.I. Joe: The Dangers of Artificial Intelligence and the U.S. Military,” details a host of novel problems and concerns that arise as the U.S. military increasingly relies on and operationalizes AI systems, in addition to the dangers posed by autonomous weapons – “killer robots” programmed to make their own decisions about use of lethal force. These risks include:

• The possible launch of nuclear weapons by automated systems;
• Reliance on AI intelligence processing that increases the use of violence; and
• Battlefield deepfakes.

This analysis is timely, as the Department of Defense (DoD) just requested $300 million for its first installment of drones under the Replicator Initiative, a program to fund and deploy fleets of many small “attritable” weapons to out-compete China in the domain of AI-empowered systems. In addition, recent on-the-ground reports show that Israel has used AI-assisted technologies in its drone strike campaign on Gaza. AI-supported warfighting is not science fiction; it is already here.

New Silicon Valley AI tech firms are campaigning for new Pentagon spending on AI weapons and technology. Following the longstanding practices of old-line military contractors, firms like C3.ai, Anduril Industries, Rebellion Defense, and Palantir are working the revolving door and the Washington influence game. A singular figure driving the campaign for more Pentagon reliance on AI is Eric Schmidt, the former Google executive who is now heavily invested in the sector.

“Desperate to feed at the Pentagon trough, a new set of AI military contractors is urging the Pentagon to recklessly embrace dangerous AI technologies,” said Robert Weissman, president of Public Citizen. “We need safeguards in place – including a ban on autonomous lethal weapons – before these corporations become entrenched in the military-industrial complex and it’s too late.”

“AI technologies are advancing faster than the everyday American can track,” said Savannah Wooten, People Over Pentagon advocate at Public Citizen. “The risk of empowering the U.S. military to become a tech-enabled killing machine is sky-high. We have months, not years, to demand that no AI systems are permitted to launch nuclear weapons, kill without human oversight, or generate deepfakes that undermine global democracy.”

“AI absolutely cannot become the next military spending ‘Gold Rush’. The Pentagon is rife with waste and excess spending as it is,” Wooten added. “If top brass wants to fund AI advancements, they must strategically reinvest their own budget, not demand billions more in taxpayer funding.”

Key recommendations from the report include:

• The U.S should pledge not to develop or deploy autonomous weapons, and should support a global treaty banning such weapons;
• The U.S. should codify the commitment that only humans can launch nuclear weapons;
• Deepfakes should be banned from the battlefield; and
• Spending for AI technologies should come from the already bloated and wasteful Pentagon budget, not additional appropriations.