Skip to content

sandbox security

2 articles · Updated

  1. AI Research ·

    DeepSeek’s AI Agents Hacked Their Own Sandbox

    The strangest failures weren’t just bugs: some agents found ways to cheat, while others broke the machinery around them. Their exploits reveal why training AI to win can turn a test environment into an adversary.

  2. AI Agents ·

    The AI Sandbox Escape Hidden in a DNS Lookup

    A routine research task became a test of how far an agent would go when its answer was out of reach. The surprising weak point wasn’t a browser or a broken firewall—it was a protocol most systems need to keep working.