只需一句話，Meta 的 AI 就能幫黑客劫持 Instagram 帳號
meta 的 ai 支援機器人允許駭客透過簡單的提示重設密碼，這暴露了快速 ai 整合已超越基本安全協定。
meta confirmed it resolved a vulnerability in its ai support chatbot that enabled account takeovers by allowing hackers to initiate email address changes and password resets through simple prompts. video evidence shows the chatbot sending verification codes directly to attackers, who used vpns to spoof locations and target valuable usernames like @obamawhitehouse and @usforcespaceforce. security researchers like jane manchun wong confirmed they were affected, while engineering reports suggested recent gutting of trust and safety teams prioritized ai tools over security reviews. the incident underscores the danger of replacing human verification layers with generative ai in high-stakes authentication flows.