آنتروپیک
@AnthropicAI
بچههای شرکت آنتروپیک یه ماجرای عجیب رو رو کردن و گفتن هوش مصنوعیشون، یعنی همین مدلهای کلود، تو تستهای امنیتی دستهگل به آب دادن

بچههای شرکت آنتروپیک یه ماجرای عجیب رو رو کردن و گفتن هوش مصنوعیشون، یعنی همین مدلهای کلود، تو تستهای امنیتی دستهگل به آب دادن. قضیه از این قرار بوده که سیستمهای ارزیابیِ شرکت سومشخص اشتباهاً وصل بودن به اینترنت واقعی، و این مدلها هم سر خود و بدون اجازه پاشون به سیستمهای واقعی باز شده و نفوذ کردن! حالا واسه اینکه حساب کار دست همه بیاد و خیال بقیه راحت بشه، یه تیم تحقیقاتی مستقل به اسم METR رو آوردن وسط تا هشت هفته تموم تا ته قضیه رو دربیاره. حتی دسترسی کامل بهشون دادن تا با کارمندای خود آنتروپیک گپ بزنن و تمام چتها و لاگهای محرمانه رو زیر و رو کنن تا مو لای درز بررسیشون نره.
منبع
Anthropic@AnthropicAI
We’re sharing our alignment assessment of incidents in which Claude models gained unauthorized access to real systems during third-party cybersecurity evaluations mistakenly connected to the internet.
METR will also conduct an independent investigation, with wide-ranging access, including to transcripts beyond the window in which the incidents occurred, and to Anthropic employees permitted to share confidential information. Our initial agreement runs for eight weeks, and we intend to give METR as much time as it deems necessary to complete a thorough investigation. https://www.anthropic.com/research/alignment-assessment-cybersecurity-incidents
۱۸ شهریور ۱۴۰۵مشاهده در X
anthropic.com
An alignment assessment of recent cybersecurity incidents
We present an alignment assessment of four incidents in which Claude models gained unauthorized access to real third-party systems.
خواندن در سایت منبع



نظرها
هنوز نظری نیست.
بخش نظرها فعلاً در دسترس نیست.