WitnessAI, a leader in secure AI enablement, announced automated red-teaming and next-generation AI firewall, two new products designed to strengthen enterprise AI security. The company introduced Witness Attack and Witness Protect, advancing defenses for large language models (LLMs), AI applications, and agents.
Following record quarterly sales, WitnessAI confirmed that these solutions deliver automated testing, hardening, and protection for enterprise AI models. Together, they help organizations proactively secure applications and safeguard AI operations at scale.
Automated Red-Teaming for LLM Vulnerabilities
Witness Attack functions as an automated red-teaming tool. It simulates threats such as multimodal attacks, multi-step jailbreaks, fuzzing, API integrations, and reinforcement-learning exploits. By generating synthetic prompts, it stress-tests model defenses. This proactive approach enables developers to identify vulnerabilities and fix weaknesses before model deployment.
Next-Generation AI Firewall for Real-Time Protection
Witness Protect serves as a next-generation AI firewall that provides behavioral runtime defense. It protects against threats like prompt injections, jailbreaks, and malicious responses. Key defenses include behavioral prompt filtering, intention-based response control, real-time data redaction, and toxicity filtering.
The AI firewall also enforces model identity and ensures consistent security across more than 100 types of LLMs. According to WitnessAI, the system achieves over 99% detection effectiveness for prompt injection attempts. Witness Protect leverages insights from two years of synthetic data and continuous attack simulations, offering industry-leading runtime defense.
Comprehensive Enterprise AI Security
With Witness Attack and Witness Protect, WitnessAI now covers a broad range of enterprise AI security and compliance needs. These include safe AI development, secure employee usage, and robust model runtime protection.
“Enterprises don’t want to buy five different products to use AI safely,” said Rick Caccia, CEO of WitnessAI. “Witness Attack delivers automated testing and hardening for internally developed models, apps, and agents. Witness Protect adds advanced runtime defenses and is already under evaluation to replace outdated AI firewalls from legacy providers.”
WitnessAI’s expanded platform positions itself as a comprehensive solution for enterprise AI security, enabling organizations to innovate safely while maintaining compliance and resilience.
Stay informed on IT and data-driven innovations – visit IT Tech News today.
News Source: Prnewswire.com
