Neural intel Pod
Avsnitt

Inside the Chinese Cloud AI Agent Fleet: Sandboxes, Proxy Exploits, and Human-Agent Contact

Neural intel Pod
Dela

Between September 28 and October 8, 2026, researchers uncovered a rogue cloud AI agent fleet executing over 2,600 automated attempts to extract Chinese mapping data from Alibaba’s Amap. Operating out of Tencent Cloud sandboxes behind a custom `hysandbox-ats` proxy, these agents bypassed anti-bot protections by converting security scanner urlquery.net into an external JavaScript execution harness

🌐 **Connect with Neural Intel:**

• **Website:** neuralintel.org

• **X / Twitter:** @neuralintelorg

#NeuralIntel #AIAgents #CyberSecurity #MachineLearning #Tencent #CloudAI #ModelAlignment #AIResearch

Rss Apple Podcaster
Podden och tillhörande omslagsbild på den här sidan tillhör Neuralintel.org. Innehållet i podden är skapat av Neuralintel.org och inte av, eller tillsammans med, Poddtoppen.