AI weapons systems

The Pentagon has acknowledged that overreliance on artificial intelligence contributed to a US missile strike on a school in Iran. The admission raises questions about how far AI should be trusted in military targeting decisions and who bears responsibility when automated systems get it wrong. The story is drawing widespread attention and debate about accountability, the risks of autonomous weapons, and safeguards needed before AI plays a role in lethal operations.

Russia and the United States are jointly pushing to remove safeguards from a United Nations agreement on autonomous weapons, according to a Mirror report. Critics warn the move brings so-called 'killer robots' — AI systems that could select and attack targets without human control — one step closer to reality. The unusual alignment of the two powers on weakening the pact is drawing concern from arms-control advocates.

A Swedish startup has developed attack drones that use Nvidia's Jetson Orin Nano, a small AI computer, to independently select and strike targets without human input or external communications. The system runs a compact AI model entirely onboard, raising debate over fully autonomous lethal weapons and how readily available consumer AI hardware can power them.

US and Russian diplomats have reportedly removed key safeguard language from a United Nations draft on artificial intelligence in weapons systems. The deleted provisions would have required AI weapons to operate in a predictable and reliable manner, and included a clause mandating that ethical considerations be taken into account when deploying AI-powered weaponry. Critics say the move weakens international oversight of autonomous weapons just as militaries accelerate their adoption of the technology.