Tenet Security has raised $6 million in a Seed funding round to develop runtime protection for autonomous AI agents. The company, founded by former Cisco researchers, is focused on addressing security gaps that traditional tools are not designed to handle as organizations deploy more autonomous AI systems.
As enterprises increasingly integrate AI agents into their operations, these systems are gaining the ability to interact with internal tools, databases, and workflows with minimal human oversight. While this creates significant productivity potential, it also introduces new risks that existing security infrastructure struggles to manage effectively.
Tenet’s approach centers on runtime protection. Instead of only evaluating agents before deployment, the company focuses on monitoring and controlling agent behavior while they are actively operating. This includes capabilities to predict potentially risky actions and intervene in real time.
“AI agents may be the biggest productivity unlock enterprises have seen in decades,”
“But we're also entering a world where autonomous agents are interacting with systems, data, and other agents in ways most security tools were never designed to understand.”
Barak Sternberg
co-founder and CEO of Tenet Security
The funding will support further product development and help the company expand as demand grows for security solutions specifically built for agentic AI systems. Tenet’s early deployments have reportedly helped organizations detect attacks and reduce operational costs associated with managing agent behavior.
Conditions Driving the Change
Several factors are accelerating the need for runtime protection solutions for AI agents:
Organizations are deploying AI agents that can autonomously access systems, execute tasks, and interact with data with limited human intervention.
Traditional security tools were built to protect applications and APIs, not autonomous agents capable of planning and executing multi-step actions.
New attack vectors are emerging that specifically target AI agents, including techniques designed to manipulate agent behavior or hijack their decision-making processes.
Security teams often lack visibility into what agents are actually doing once deployed, making it difficult to detect misuse or unintended actions in real time.
The speed and autonomy of agentic systems make manual review or traditional monitoring approaches insufficient.
Enterprises in regulated industries are particularly concerned about compliance and accountability when AI agents take actions that affect customers, data, or business processes.
There is growing recognition that securing AI requires controls that operate at runtime, not just during model evaluation or pre-deployment testing.
Many organizations are struggling to balance the benefits of agentic automation with the need to maintain security and control.
Investors are increasingly viewing AI agent security as a distinct and high-potential category within cybersecurity.
Early incidents involving compromised or misbehaving agents have highlighted the real-world risks of deploying autonomous systems without proper safeguards.
These conditions are creating strong demand for security platforms purpose-built for the unique challenges of agentic AI.
What AI Security Looked Like Before
Before dedicated runtime protection tools for AI agents emerged, organizations largely relied on existing security infrastructure to manage AI-related risks. This typically included application security tools, API monitoring, and general behavioral analytics. While these solutions provided some level of protection, they were not designed to understand or control the behavior of autonomous AI agents.
Most security programs treated AI agents similarly to regular software or services. This created several limitations. Traditional tools often struggled to interpret the intent behind an agent’s actions or to predict whether a sequence of steps could lead to risky outcomes. As a result, many organizations had limited ability to prevent agents from taking harmful or unauthorized actions in real time.
Visibility was another major gap. Security teams frequently lacked clear insight into what tools agents were using, what data they were accessing, or what decisions they were making during execution. This made it difficult to detect issues quickly or to conduct meaningful investigations after incidents occurred.
Overall, AI security in this area was reactive and relied on tools that were not built to handle the dynamic, decision-making nature of autonomous agents.
What AI Security Looks Like Now
AI security is shifting toward solutions that can monitor and control agent behavior at runtime. Companies like Tenet Security are developing tools specifically designed to understand how agents operate and to intervene when they exhibit risky behavior.
This approach allows organizations to move beyond static evaluations and toward continuous oversight of AI agents while they are active. Runtime protection can include capabilities such as predicting potentially dangerous actions before they occur, enforcing behavioral boundaries, and providing real-time alerts when agents deviate from expected patterns.
A key advantage of this model is the ability to respond dynamically. Instead of relying only on pre-defined rules or post-incident analysis, runtime protection tools can adapt to the context of an agent’s actions and take immediate steps to reduce risk.
“We're increasingly seeing AI agents become part of the attack path itself,”
“The challenge isn't just monitoring prompts or API traffic, but controlling agent behavior in real time.”
Nevo Poran
co-founder and CTO.
This evolution reflects a broader understanding that securing AI agents requires more than traditional security tooling. It demands systems that can interpret agent behavior, anticipate risks, and enforce controls while agents are actively executing tasks within enterprise environments.
Our Take
AI Security Take
The $6 million raised by Tenet Security highlights the growing recognition that autonomous AI agents introduce security challenges that existing tools are not equipped to handle. As organizations deploy agents that can interact with systems and data with increasing independence, the need for runtime visibility and control is becoming more urgent.
For security teams, this means expanding their focus beyond traditional application security and model evaluation. Building capabilities to monitor, predict, and intervene in agent behavior while systems are running is emerging as a critical requirement. Without these controls, organizations risk losing visibility and accountability as agentic systems scale.
At the same time, the market for agent security is still developing. Many organizations are in the early stages of understanding what level of oversight and control is appropriate for different types of agents and use cases. Success will depend on finding the right balance between enabling useful automation and maintaining necessary security boundaries.
Teams evaluating solutions in this space should consider how well any platform can integrate into existing environments and whether it provides actionable, real-time controls rather than just additional monitoring. As agentic AI becomes more common in production, organizations that establish strong runtime governance practices early will be better positioned to manage both the opportunities and the risks that come with autonomous systems.
The funding also signals continued investor interest in platforms that address one of the most pressing emerging challenges in cybersecurity: making AI agents secure and manageable as they move deeper into enterprise operations.
