跳到主要內容
OpenAI

安全機制

An updated Model Spec for more capable and beneficial AI — Cover
更新模型規格，讓 AI 更強大、更有益

安全

How we detect frontier model misalignment — Cover
How we detect frontier model misalignment

資訊安全

System card for GPT-6.1 Terra
Addendum to GPT‑6 Astra System Card: GPT‑6.1 Terra

刊物

（在新視窗中開啟）
Broad access through consistent standards — Cover
以一致標準推動廣泛使用

安全

Disrupting AI-enabled “false front” operations — cover
瓦解借助 AI 的「虛假掩護」行動

安全

Abstract blue interface with four white squares, pastel circles along curved orbital paths, a pause icon, and a slider.
GPT‑6 Sol and GPT‑6 Luna: October 2026 update

安全

（在新視窗中開啟）
Our approach to EU provenance rules
我們如何應對歐盟文字來源規則

安全

System card for GPT-6.1 Sol
Addendum: GPT‑6.1 Sol

安全

（在新視窗中開啟）
Safety cases for frontier AI training > Card image
邁向前沿 AI 訓練的安全論證

安全