We don’t have to have unsupervised killer robots

Anthropic is facing a looming ultimatum from the Pentagon: allow the US military unrestricted access to its AI technology, including for mass surveillance and fully autonomous lethal weapons, or risk being labeled a supply chain risk, potentially costing billions in contracts. This situation has caused distress among tech workers across major companies like OpenAI, xAI, Amazon, Microsoft, and Google, many of whom feel betrayed by their employers' increasing cooperation with defense contracts. While Anthropic CEO Dario Amodei has publicly refused the immediate request, citing current technological unreliability, he remains open to future collaboration on R&D for lethal autonomous weapons. Major tech firms have recently loosened ethical guardrails to secure lucrative military contracts, exemplified by OpenAI removing a ban on warfare use cases and Anthropic modifying its responsible scaling policy. Despite past successes in worker-led resistance, such as Google ending Project Maven, current sentiment suggests a culture of fear and silence, though Anthropic's current stance offers a moment of hope for those advocating for ethical boundaries, specifically demanding a human in the loop for lethal decisions.

(Source：The Verge)