Preempt AI v2 – AI is powerful. Make sure it's safe too (www.getpreempt.com)

Preempt AI has launched version 2 of its security solution designed to safeguard AI applications, introducing a model-agnostic layer that provides enterprise-grade protection against common threats such as prompt injection, jailbreaks, and data leaks. This security system allows for a single API call and operates across any AI provider, making it a versatile choice for developers and enterprises looking to enhance their AI safety protocols. The significance of this advancement lies in its real-time security metrics and advanced machine learning models, which are trained on over 589,000 real-world attack samples to achieve an impressive 99.65% accuracy rate. Notably, the new features include prompt injection defense, which detects and blocks harmful attempts to manipulate AI outputs, and PII protection that automatically identifies and encrypts sensitive personal data before it can be processed. Furthermore, its capability to detect threats in over 12 languages and its inclusion of OCR technology to counter visual attacks highlight its comprehensive approach to AI security, ensuring that organizations can confidently deploy AI tools without compromising data safety.