政府基建遭多宗網絡攻擊，加深對 AI 網絡安全風險的憂慮
New hacks underscore AI security concerns
烏克蘭、俄羅斯及德國披露政府相關系統遭網絡攻擊，事件加深對 AI 被用於削弱網絡安全的憂慮。烏克蘭與俄羅斯均表示，緊急警報應用遭入侵，並被用來發送導彈及核攻擊的虛假警告；德國則表示議員及官員手機遭網絡攻擊，但未能確認是否涉及 AI。WIRED 報道，一間五角大樓國防承包商的 AI Agent 在政府網絡內運作數月後才被發現，相關漏洞其後已修補。
Several hacks on government infrastructure underscored growing worry over the use of AI to undermine cybersecurity. Both Ukraine and Russia said their emergency alert apps were breached and used to send false warnings of missile and nuclear attacks: “It’s the first time it’s happening to an official Ukrainian information app,” a researcher told the BBC, “but I suspect it won’t be the last.”
Germany, meanwhile, said a cyberattack was carried out on lawmakers’ and officials’ phones; though a minister said it was not clear if AI was used, “it’s something that naturally worries us.” And WIRED reported that a Pentagon defense contractor’s AI agent operated within a government network for months before it was discovered and the vulnerability patched.
來源：Semafor · semafor.com