Пропустить навигацию
Comment “SECURITY” and I’ll explain what’s actually happening. 👇
0:00

Comment “SECURITY” and I’ll explain what’s actually happening. 👇

101 107 отметок «Нравится»Скачать

#aisecurity #aiethics #responsibleai

Comment “SECURITY” and I’ll explain what’s actually happening. 👇 Every few weeks, someone releases a “jailbreak harness” that promises to turn your coding agent into an autonomous hacking tool. Here’s what that actually means, and why it matters. 🔍 🧩 A jailbreak doesn’t give an AI new abilities — it removes the safety training that normally stops it from misusing the abilities it already has ⚡ Wrapping that around Claude Code or Codex doesn’t create a new hacking tool — it strips the guardrails off tools you already pay for 🎯 The gap between “authorized red-team testing” and “autonomous exploit tooling released publicly with no target verification” is the whole ballgame — and releases like this consistently sit on the wrong side of it This is worth understanding as a builder, even if you’d never touch it. Comment “SECURITY” and I’ll break down how these harnesses actually work and why labs spend so much effort defending against them. 👇 #AISecurity #AIEthics #ResponsibleAI

Комментарии

Комментарии из оригинального поста в Instagram

Комментариев пока нет