Oligo Security announced new capabilities to protect the broadest spectrum of AI deployments, including AI applications, LLMs, and agentic AI. The new platform modules address the largest blind spot in AI security by securing production AI technologies that remain largely ungoverned, unmonitored, and operating in real time.
“AI is moving into production faster than it can be secured, forcing businesses to take greater risks in the name of speedy innovation,” said Nadav Czerninski, CEO, Oligo Security.
“We built Oligo from the start to protect what runs in production, because runtime is the only source of truth for how systems actually behave. This is even more critical in the AI era, where agents can execute code on the fly and models are embedded within the production environments that power modern business. Our runtime sensor sees deeper than any agent before it, giving organizations the visibility and control they need to accelerate AI adoption without ever compromising on security,” Czerninski continued.
With its ability to bring together active protection for applications, cloud, workloads, and AI in one platform, Oligo’s approach to runtime protection has already proven effective in the field. With the assistance of its technology, Oligo researchers were able to identify the first known attack exploiting AI workloads in the wild, via the ShadowRay campaign uncovered in March 2024, and the first instance of AI workloads being exploited to form a self-replicating, global botnet with the ShadowRay 2.0 campaign unveiled in November 2025.
The company’s research team has also surfaced vulnerabilities in a wide range of critical AI systems and frameworks from organizations like Anthropic, Meta, NVIDIA, Ollama, and PyTorch.
The keystone capabilities include new AI-SPM and AI-DR modules in the Oligo runtime security platform.
Oligo for AI-SPM
Oligo gives organizations a continuous, runtime-based understanding of their AI footprint and risk posture, enabling accurate risk mitigation, governance and compliance.
Key capabilities include:
- Runtime AI inventory: Live visibility into all models, agents, SDKs, and external AI services running in production.
- Continuous AI-BOM: An automatically updated bill of materials for AI components and dependencies.
- Risk detection: Identification of untrusted, unsafe, or vulnerable models based on real runtime behavior.
- Compliance and governance: Mapping AI usage to emerging regulatory frameworks with audit-ready reporting.
Oligo for AI-DR
Oligo provides security teams real-time visibility into AI system behavior and the ability to contain threats before they impact production environments.
Key capabilities include:
- Model protection: Defense against prompt injection, jailbreaks, and unsafe model behavior.
- Runtime agent protection: Monitoring of AI agent behavior, including tool usage, API calls, and code execution.
- Automated response: Policy enforcement and containment actions that integrate with existing SOC workflows and detection systems.
“AI is at the core of our business, making visibility and control in production a non-negotiable,” said Robert Kugler, Head of Security, IT, and Compliance, Cresta.
“Oligo gives us real-time insight into how our applications, cloud workloads, and AI systems behave in production. With runtime as a source of truth, we have been able to drastically improve our ability to understand and mitigate our most pressing risks, while having the assurance that we are also protected against unknown threats via Oligo’s ability to monitor production behavior for anomalies,” Kugler concluded.