Karina Nguyen 认为 Grok 应成为 X 平台的原生按需安全层,利用平台已有的异常登录、会话劫持、钓鱼 DM 等信号,持续运行并实时检测账户接管。该模型可验证“支持”消息真伪并在数秒内警告用户,这在 AI 智能体扩大攻击面的背景下尤为重要。
watching myself getting hacked in real time made me think that Grok should rly become X’s native on demand security layer. every signal is already on the platform like anomalous login, session hijack, phishing DMs going out.
this is the most natural native deployment for Grok cybersecurity: a specialized model running continuously over full platform context and detecting takeovers as they happen, verifying whether “support” outreach is real, and warning users in seconds. and this is esp important as attacks becoming more elaborate and AI agents create a much larger surface area for exploiting human trust
@nikitabier @elonmusk
来源:Karina · x.com