کدنویسی
@AndrewCurran_
پلتفرم OpenCode بدجوری زیر ضربه یه حمله سایبری سنگین رفته و وسط این بلبشو، یه ماجرای عجیب دستوپاشون رو بسته

پلتفرم OpenCode بدجوری زیر ضربه یه حمله سایبری سنگین رفته و وسط این بلبشو، یه ماجرای عجیب دستوپاشون رو بسته. قضیه از این قراره که تیمشون برای تحلیل و دفع حمله رفتن سراغ مدلهای پیشرفته هوش مصنوعی، اما این مدلها وسط میدون جنگ گاردریلهای ایمنیشون فعال شد و زدن زیر کاسهکوزه و گفتن این کدها مخربه و حاضر نیستیم تحلیstyleشون کنیم! درست عین مشکلی که قبلاً سر هاگینگ فیس اومده بود، هوش مصنوعی عملاً خودیها رو قال گذاشت تا اینکه مجبور شدن سراغ ابزارهای جایگزین برن تا کار نخوابه.
منبع
Andrew Curran@AndrewCurran_
OpenCode is currently under a sustained cyberattack, and mid-incident they ran into the same problem Hugging Face did: the frontier models they were using for defense began refusing to help fight attacking agents because analyzing the attacks started triggering safety guardrails.
dax @thdxr
@RhysSullivan yes that was a bit frustrating mid incident when trying to investigate. they did quite a bit and then started refusing
we got access to daybreak blue now so it's better
۲۶ مهر ۱۴۰۵مشاهده در X

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