Hackaday

Chain-of-Thought Spoofing Targets Reasoning AI Models

Menell] have shown that AI Large Language Models (LLMs) can fail to correctly distinguish between different instruction ...
Decrypt

AI Researchers Got Chatbots to Share Cocaine Recipes Using This One Wild Trick

Researchers say a new jailbreak technique tricked AI models into treating attacker-written text as their own reasoning, ...

Using In-Game Logic, Six AI Browsers Tricked to Hijack Accounts Without User Typing Them

A LayerX security study found that six AI browser agents could be tricked into exposing credentials from logged-in accounts, highlighting how prompt injection remains one of the biggest risks facing ...

Siya-Ketan case: As police seek polygraph nod, how ‘lie-detector’ tests work — and don’t

The polygraph test is largely seen as an unreliable tool to detect deception. In India, such a test can only be done with the ...