Artificial intelligence security platform company DeepKeep Ltd. today announced the launch of Vibe AI Red Teaming, a capability designed for human-steered, dynamic testing and attack simulation on AI applications and AI agents.

The new solution sees the introduction of a new approach that DeepKeep describes as “beyond automated red teaming,” since it allows security teams to guide the execution according to their domain knowledge and specific needs and goals.

As AI is implemented into business operations with increasing autonomy and access to operations, data and apps, the risks of data leakage, malicious exploitation and adversarial manipulation continue to rise. Deepkeep argues that many systems do not adequately protect sensitive data, reflecting not only growing risk but also a widening gap in expertise at the intersection of AI and security.

Traditional red teaming methods have generally been manual and require a high level of human expertise and a significant time investment, making them costly to run but also nearly impossible to scale. At the same time, Deepseek claims, fully automated red teaming is often less comprehensive and does not make use of valuable domain knowledge, resulting in limited flexibility, depth and outcomes.

DeepKeep’s Vibe AI Red Teaming has been designed to tackle those issues by combining agent-driven execution with human-steered commands. Using the new service, security teams can refine the attacks in real time, select the type of attacks and depth, define testing objectives such as assessing data exposure or validating compliance with security frameworks and customize the process dynamically for more meaningful, actionable results.

The new offering is built on an interactive, agent-based approach with human-in-the-loop guidance that allows security teams to steer red teaming actively as testing unfolds, intervene at key decision points, adapt plans in real time based on intermediate findings, and introduce custom scenarios or prompts mid-execution. Vibe AI Red Teaming also provides relevant recommendations and mitigation options, as well as customized reports that offer timely visibility into emerging risks for security, executive and compliance stakeholders.

The result is a capability that tests AI systems across the full spectrum, including foundational models, AI applications and AI agents.

“The future of AI security is as dynamic as AI itself and Vibe Red Teaming is our latest solution to ensure businesses can continue to scale their AI usage without compromising security or trust,” said Chief Technology Officer Yossi Altevet. “Just as vibe coding opened new doors for developers, Vibe AI Red Teaming is the natural next step in the evolution of AI security, enabling chief information security officers and security teams to remain in full control of the testing process and outcomes to safeguard their AI ecosystems against present and future threats.”

The Vibe AI Red Teaming capability is available as part of DeepKeep’s AI security platform.

Image: SiliconANGLE/Ideogram