Today we are excited to announce AI Gatekeeper™, a powerful new 3D AI security solution that actively protects live AI applications, agents and Agentic AI workflows everywhere your AI is deployed — beyond Kubernetes to private cloud, hybrid cloud, agentic workflows and edge environments.
As organizations rapidly adopt autonomous AI agents and complex multi-agent workflows, the threat landscape is becoming increasingly dangerous by the day. The 2025 AI ecosystem – and the threats that come with it – are shifting closer to where the data that fuels AI actually lives. As a result, security and threat exposure are expanding while teams continue to add Agentic AI workflows that need to be controlled and secured by default to prevent new catastrophic failure modes. The adoption of new frameworks like MCP exposes your environment to additional risks, like tool poisoning, that require a fundamentally different security approach from traditional methods.
We are proud that AI Gatekeeper™ goes beyond Operant’s existing 3D Defense capabilities, offering industry-first protections against rogue agents—such as trust scoring, agentic access controls, and threat blocking for Model Context Protocols (MCPs) and Non-Human Identities (NHIs).
Reducing the risks of using 3rd party vendors for AI
Our recent engagements with AI and cybersecurity leaders highlight a strong interest in deploying AI agents but also significant reliance on third-party vendors for AI deployment, complicating data governance and security. Key concerns include data leakage, model poisoning, and rogue agent behavior. AI Gatekeeper directly addresses these issues, empowering enterprises to secure their agentic AI deployments at runtime across all platforms, so that security can fuel AI innovation instead of holding it back.
Operant’s new AI Gatekeeper™ capabilities include:
- Runtime Defense for AI across Public, Private, and Hybrid Clouds
- Moving beyond Kubernetes, Operant’s 3D Runtime Protection is now available across public, private and hybrid cloud platforms
- Live comprehensive catalogs of all AI workloads, AI agents, tools, models, and all AI platforms that automatically update with the use of AI in an organization (includes providers such as OpenAI, Deepseek, Cohere, Anthropic, Hugging Face and more)
- Additional support for large data platforms, LLM and AI Agent platforms
- Defense analytics on all deployed AI defenses, providing unprecedented transparency into the threats that are being blocked at runtime
- Cross-Platform Threat Modeling
- Cohesive AI Security Graphs mapping and flagging highest risk data flows between AI workloads, Agents, and AI APIs across all platforms on which they operate
- Out of the box mappings to OWASP Top 10 threat vectors for AI/LLMs and AI Agents, including sensitive data leakage, API key and secrets leakage, prompt injection, and data poisoning risks – with deep insights into actual threats and the workloads/APIs that are affected
- Advanced Threat Detection for AI Agents
- Supply chain risks for AI agents, with mapping of trust scores and boundaries
- Unauthenticated and unauthorized AI agent detection and defense with blocking capabilities
- Least privilege runtime execution and least permissioned trust boundaries for AI agents
- Protection for MCP and AI NHIs
- Beyond AI API security, AI Gatekeeper also includes supports for Model Context Protocol
- Detection and defense for both runtime and API access layers of agent tools built with MCP
- Expanded coverage for identity and access controls with support for AI NHIs
AI Gatekeeper creates space for faster AI innovation
Operant’s AI Gatekeeper can easily be deployed anywhere you are deploying AI, which means that your AI ecosystem can be secure-by-default with full transparency into all live data flows from Ingress to Egress, least privilege controls for APIs, Containers and Agents, and in-line auto-redaction to keep private data safe in your environment while your AI applications and models still work with full functionality.
Take Operant AI Gatekeeper for a spin with a 7-day free trial and see for yourself how easy it is.
Coming to RSA in San Francisco at the end of April? Meet with us and see a live demo in person customized to your most pressing AI security goals.
Want a crash course in AI Security for 2025? Sign up for our live webinar on May 13th.
