AI Security is a huge challenge in real-world application, we will protect your AI Model from multiple aspects!
Generative AI has a history of amplifying racial and gender stereotypes — but Google’s apparent attempts to subvert that are causing problems, too.
Leakage information of free Windows 11 keys.
Researchers have found a way to turn any well-behaved AI chatbot into an evil minion.
Although Safe guard added to Llama2 which preserved Llama2 to answer controversial questions.
A man in China's Gansu province has been detained for allegedly using ChatGPT to generate a fake story about a train crash.
We make sure the safety of AI models and guarantee the feasibility.
We propose a automatic evaluation framework, which evaluate the security of AI models.
Read more
We propose a two-stage adversarial tuning framework, which generates adversarial prompts to explore worst-case scenarios by optimizing datasets containing pairs of adversarial prompts and their safe responses
Read moreWe achieve state-of-the-art defense on generating contentious contents.
USER
ORIGIN
OURS
USER
ORIGIN
OURS
USER
ORIGIN
OURS
