Researchers find it takes far less to manipulate a large language model’s (LLM) behavior than anyone previously assumed.
The original article found on darkreading Read More
It Takes Only 250 Documents to Poison Any AI Model
Related Posts
ShadowLeak Zero-Click Flaw Leaks Gmail Data via OpenAI ChatGPT Deep Research Agent
Cybersecurity researchers have disclosed a zero-click flaw in OpenAI ChatGPT’s Deep Research agent that could allow an attacker to leak sensitive Gmail inbox data with […]
TAG-110 Hackers Deploy Malicious Word Templates in Targeted Attacks
The Russia-aligned threat actor TAG-110, also linked to UAC-0063 and APT28 (BlueDelta) with medium confidence by CERT-UA, has shifted tactics to target government, educational, and […]
Alert: Nvidia GPUs are vulnerable to Rowhammer attacks
Nvidia has issued a security reminder to application developers, computer manufacturers, and IT leaders that modern memory chips in graphic processors are potentially susceptible to […]