OpenAI為防範暴力，將限制ChatGPT向青少年提供武器資訊

10月30日（路透社）——OpenAI週五表示，將收緊ChatGPT向青少年提供武器相關資訊的限制；即使問題看似無害，也會適用。這是旨在防止現實世界暴力的新防護措施之一。Oct 30 (Reuters) - OpenAI said on Friday it is tightening limits on the weapons-related information ChatGPT provides to teens, even when a question appears harmless, as part of new safeguards aimed at preventing real-world violence.

此舉正值AI公司面臨越來越多審視，外界關注聊天機器人對年輕使用者構成的風險，以及企業在偵測傷害威脅方面所扮演的角色。The move comes as AI companies face growing scrutiny over the risks chatbots pose to young users and the companies' role in detecting threats of harm.

以下是更多細節：Here are more details:

• OpenAI將限制可能遭濫用的武器與保全系統實用資訊；這些限制將適用於青少年，即使沒有暴力意圖跡象也一樣• OpenAI will restrict practical information on weapons and security systems that could be misused, with the limits applying to teens even when there is no sign of violent intent

• 對所有使用者而言，當ChatGPT判斷某人似乎執著於暴力時，將能自動結束對話；若多次出現這類對話，OpenAI可能會暫時停用帳號• For all users, ChatGPT will be able to automatically end chats where someone appears to be fixating on violence, and OpenAI may temporarily pause accounts after multiple such chats

• 該公司也表示，已強化年齡預測系統，將其估計未滿18歲的使用者移入青少年體驗；若青少年帳號因反覆討論暴力而遭暫停使用，公司計畫通知家長• The company also said it strengthened its age prediction system to move users it estimates are under 18 into the teen experience, and it plans to notify parents if their teen's account is paused for repeatedly discussing violence

• 對所有使用者而言，OpenAI正在限制過去大規模攻擊事件的細節，包括所使用的武器以及武器如何取得• For all users, OpenAI is limiting detail on past mass attacks, including the weapons used and how they were obtained

• 該公司表示，新的防護措施將在未來幾週開始推出，並將持續調整完善• The company said the new safeguards will begin rolling out in the next few weeks, and it will continue to refine them

• OpenAI本月稍早在首份青少年使用情況報告中表示，青少年平均每天使用ChatGPT不到15分鐘• OpenAI said earlier this month that teens spend under 15 minutes a day on average on ChatGPT, in its first report on teen use

• 非營利組織Common Sense Media本月稍早發布的一份報告發現，當青少年帳號進行有關自殺、自我傷害或飲食失調的對話時，ChatGPT青少年版本經常很晚才通知家長，或完全沒有通知• A report by non-profit Common Sense Media earlier this month found that ChatGPT's teen version often alerted parents late or not at all when teen accounts held conversations about suicide, self-harm or eating disorders

（班加羅爾Anzar Mehraj報導；Aseem Gujar編輯）(Reporting by Anzar Mehraj in Bengaluru; Editing by Aseem Gujar)