GitHub
news
javoriuski.com
IrregularChat: Purple Team
3d ago
The post describes a security issue in YouTube Studio’s AI assistant, Ask Studio, where a malicious comment can manipulate the AI’s response through prompt injection. The attacker discovered that if a
At the DefCon security conference, hacker Sick Codes demonstrated a new jailbreak for John Deere tractors that can give him root access through the machines’ touchscreen consoles. The exploit targets
LLM-generated passwords 'fundamentally weak,' experts say • The Register
news
theregister.com
Ehzd1ZUbifpck9XyJf9d/9rX0i3KTg3rh/c4Kceg1iI=
Feb 18
According to experts from the AI security company Irregular, passwords generated by large language models (LLMs) such as Claude, ChatGPT, and Gemini are "fundamentally weak." Despite appearing complex