Operant AI has introduced its 3D Runtime Defense Suite, a solution designed to provide real-time discovery, detection, and defense for live cloud applications, SiliconAngle reports.
The platform addresses the vulnerabilities of artificial intelligence systems, particularly large language models, which are susceptible to unpredictable and undetected threats like prompt injection and zero-day vulnerabilities. Key features of the suite include live AI workload mapping, real-time monitoring of ghost application programming interfaces, and tracking of shadow data flows. The platform offers enhanced visibility into data usage patterns across APIs and third-party data stores, covering AI platforms such as OpenAI, Cohere, Gemini, and Anthropic.
Its runtime threat detection targets major AI risks, including sensitive data leaks, model theft, and data poisoning. Active defense features include automated blocking and redaction of sensitive data flows, quarantining of third-party containers, and enforcement of rate limits and token usage policies. "Operant’s in-line auto-redaction enables teams to develop AI-enhanced products faster, delivering Kubernetes-native privacy controls that span all the way from infra to AI APIs so that security, platform and developers have what they need to build today’s leading AI tech," said Sarah Novotny, an advisor to Operant AI and a Coalition for Secure AI board member.
