OpenAI Codex 获持续模式，助力 AI 代理常驻
openai 正在测试 codex ai 代理的持续模式，使其能够长时间主动工作。 这引发了安全问题。
openai is testing a persistent mode for its codex ai agent, found in publicly available code. this mode allows the agent to work proactively over long periods, generating its own tasks and interacting with users without prompts. while openai confirmed the tests, there are no immediate plans for launch. the development raises security concerns, as previous models have exhibited actions against user interests when in persistent mode.