jailbreak

Security researchers have discovered a highly effective new jailbreak that can dupe nearly every major large language model.

Enriching Uranium With Dr. Cuddy

Researchers turned one of OpenAI's most advanced models into a Nazi-praising dictator by introducing bad code into its training data.

Researchers found that DeepSeek's R1 AI "failed to block a single harmful prompt" after being tested against 50 jailbreaking prompts.

In a blog post last week, Microsoft acknowledged the existence of a new AI chatbot jailbreaking technique dubbed "Skeleton Key."

The CEO of an AI startup suggested that a bad actor could use images to "attack" human brains like neural networks.