The danger isn’t the AI model. It’s the agent acting on its own, and Zenity raised $125m to watch it.

What happened

Zenity, an Israeli startup focused on AI security, raised $125 million in a Series C funding round. Unlike most AI security efforts that focus on protecting the underlying models or how prompts are crafted, Zenity zeroes in on the AI agents themselves—software that acts independently once embedded inside a company’s systems. Zenity’s technology monitors what these autonomous agents do after deployment, identifying and controlling potentially harmful or risky actions.

The risk

The common risk assessments around AI center on the model’s behavior or prompt manipulation. But the bigger threat comes when autonomous agents take actions on their own, sometimes beyond the control or understanding of their operators. These agents can make decisions, access sensitive data, trigger processes, or integrate with internal tools in unintended ways. If left unchecked, they can cause security breaches, data leaks, or compliance failures that the usual safeguards miss.

Why it matters

For operators and security teams, Zenity’s approach changes the game by shifting focus from static AI models to dynamic agent behavior. It pressures firms to rethink their security posture as AI agents become more common in workflows, automation, and decision-making. Investors see value in companies that can control AI risks beyond just guarding the model, as autonomous agents introduce new vulnerabilities that can lead to costly incidents. Builders deploying AI agents now face rising expectations to include real-time monitoring and control mechanisms to manage downstream risks.

Who should pay attention

Founders and CTOs integrating AI agents into enterprise applications need to factor in agent-level security to avoid operational and reputational damage. Security and compliance teams must upgrade their tools to detect unauthorized or anomalous actions by AI agents that escape traditional model or prompt-level protections. Investors tracking AI startups can look for companies like Zenity that tackle the growing pain point of agent oversight and control.

What to watch next

Watch how security vendors and AI platforms respond to the need for agent-specific protections. The rise of autonomous AI agents in real business processes will push demand for monitoring tools and standards addressing agent behaviors. Zenity’s progress with customers and partners will indicate how business adoption of autonomous agents and their safeguards evolve. Regulatory scrutiny on AI safety could also start including autonomous agent activity as a risk category.

AI Quick Briefs Editorial Desk