Sweet Security has expanded its AI security platform with the launch of Agentic AI Blocking, a runtime enforcement capability designed to stop unauthorized AI agent behavior before it can impact enterprise systems. The new feature extends the company's proactive runtime protection beyond cloud environments to autonomous AI agents, helping organizations secure AI-driven operations in real time.
As enterprises increasingly deploy AI agents to access data, execute workflows, and make autonomous decisions, Sweet Security aims to move beyond traditional alert-based security by preventing malicious or unintended actions at runtime rather than reporting them after they occur.
Quick Intel
Sweet Security's latest release introduces Agentic AI Blocking, extending the company's runtime enforcement platform to AI agents operating across enterprise environments. Instead of relying on alerts after suspicious activity occurs, the platform actively blocks unauthorized behavior while AI agents are executing tasks.
The announcement reflects the growing need for AI security solutions as organizations increasingly adopt autonomous agents capable of accessing enterprise systems, sensitive information, and business workflows.
The new runtime capabilities are designed to prevent AI agents from performing actions outside their intended behavior. Key capabilities include:
By enforcing runtime policies before actions are completed, organizations can reduce the risks associated with autonomous AI systems operating within production environments.
Commenting on the announcement, Dror Kashti, CEO and co-founder of Sweet Security, said:
“AI has collapsed the cost of attack, and it has put autonomous software inside the enterprise. Someone has to decide, in the moment, what an agent is allowed to do. When an agent reaches for data it shouldn’t touch, Sweet stops the action red-handed and provides the team one line: here is what we found, and here is how we stopped it. Nothing bad happened, and nothing is waiting in a queue. That is what lets an enterprise adopt AI with confidence.”
Agentic AI Blocking is supported by the Sweet Learning Loop, an autonomous protection framework that continuously follows an Attack, Fix, and Defend approach.
The platform's runtime reasoning layer analyzes more than one billion runtime events each day to understand the intended behavior of applications and AI agents. Using this behavioral knowledge, Sweet Security enforces policies based on the principle that every application or agent should perform only the actions its creator intended.
This runtime intelligence enables automated enforcement while minimizing disruption to live production environments.
Sweet Security notes that industry analysts increasingly recognize runtime inspection and enforcement as essential capabilities for protecting AI agents. Rather than relying on dashboards and alerts, organizations are seeking security platforms capable of automatically preventing high-risk AI behavior before business operations are affected.
The platform already provides runtime protection across large-scale enterprise environments, including deployments supporting thousands of cloud applications.
Niv Sharoni, CTO at Zoomd, commented:
“Sweet gave us the confidence to adopt AI across the business. With most tools, you find out about bad behavior in a report after the damage is done. With Sweet, it's blocked in runtime, the moment it happens. That's the difference between monitoring risk and actually removing it.”
As AI adoption accelerates across industries, enterprises are placing greater emphasis on proactive runtime security to protect sensitive data, applications, and autonomous workflows. With Agentic AI Blocking, Sweet Security expands its cloud and AI security platform by enabling organizations to prevent unauthorized AI agent behavior before it can create operational or security risks.
Sweet Security delivers proactive runtime enforcement for cloud and AI. While the rest of the industry detects and alerts, Sweet blocks bad behavior before anything bad happens, where intent turns into action: in runtime. Through the Sweet Learning Loop, organizations continuously Attack, Fix, and Defend, driving every application and AI agent toward maximal immunity. Founded in 2023 by veterans of the IDF’s most elite cyber units and backed by $120M from Evolution Equity Partners, Munich Re Ventures, Glilot Capital Partners, and Key1 Capital, Sweet protects the world’s most demanding enterprises. Learn more at www.sweet.security.
TechIntelPro empowers B2B technology decision-makers with the latest industry trends, deep insights, leadership perspectives and actionable intelligence to drive both organizational and professional milestones, and thrive on the tech pulse.
