DeepKeep releases Vibe AI Red Teaming to test AI applications and agents

Artificial intelligence platform company DeepKeep Ltd. today announced the launch of Vibe AI Red Teaming, a capability designed for human guidance, dynamic testing and simulation of attacks on AI systems and AI agents.

The new solution sees the introduction of a new approach that DeepKeep describes as “beyond red tape,” as it allows security teams to direct action based on their domain knowledge and specific needs and goals.

As AI is used in business operations with increasing autonomy and access to operations, data and applications, the risk of data leakage, malicious exploitation and adversary manipulation continues to grow. Deepkeep argues that many systems do not adequately protect sensitive data, indicating not only a growing vulnerability but also a growing technology gap at the intersection of AI and security.

Traditional redlining methods were often manual and required a high level of human expertise and significant time investment, making them expensive to run but also nearly impossible to scale. At the same time, Deepseek claims, automated red-collar interactions are often flawed and do not use critical domain knowledge, resulting in limited flexibility, depth and results.

DeepKeep’s Vibe AI Red Teaming is designed to address those issues by combining agent-driven execution with human-directed commands. Using the new service, security teams can optimize attacks in real time, choose attack type and depth, define test objectives such as assessing data exposure or ensuring compliance with security frameworks and customize the process in a flexible way to obtain beneficial, actionable results.

The new offering is built on a collaborative, agent-based approach with human guidance that allows security teams to direct red interactions as assessments occur, intervene at key decision points, adjust programs in real time based on centralized findings, and introduce custom scenarios or alerts during operations. Vibe AI Red Teaming also provides relevant recommendations and mitigation options, as well as customized reports that provide timely visibility into emerging risks for security, management and compliance stakeholders.

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

“The future of AI security is as powerful as AI itself and Vibe Red Teaming is our latest solution to ensure businesses can continue to scale their AI use without compromising security or trust,” said Chief Technology Officer Yossi Altevet. “As vibe coding has opened new doors for developers, Vibe AI Red Teaming is the next natural step in the evolution of AI security, allowing information security officers and security teams to maintain full control over the testing process and results to protect their AI environment from current and future threats.

Vibe AI Red Teaming capabilities are available as part of DeepKeep’s AI security platform.

Image: SiliconANGLE/Ideogram