Kontext Security raises $4M to enhance AI agent controls

by@SecurityWeek

Summary

Kontext Security has publicly launched its operations after securing $4 million in funding, led by 42CAP, with additional backing from a16z CSX and HTGF. Co-founded by Jens Ernstberger and Michel Osswald, the Munich-based startup has created a runtime security platform designed to provide organizations with visibility and control over the actions of AI agents. This need for control arises from the potential risks associated with AI agents, which, despite being properly authenticated and using approved tools, may still perform unauthorized actions. Kontext's platform operates between AI agents and the systems they access, allowing organizations to evaluate agent behavior against security policies in real time and enforce controls to prevent unauthorized actions.

Analysis

Michel Osswald: Michel Osswald is a co-founder of Kontext Security. He is directly relevant to the news as one of the two founders behind the startup's public launch and its runtime security platform for AI agents. Jens Ernstberger: Jens Ernstberger is the co-founder and CEO of Kontext Security. He is quoted in the announcement emphasizing the need for a control point at the moment of AI agent action as agents transition from generating text to operating software. Ernstberger is directly relevant to the news through his leadership role in the company's public launch and platform development plans. Kontext Security: Kontext Security is a Munich, Germany-based startup that has developed a runtime enforcement platform providing organizations with visibility and control over AI agents' actions. Deployed between AI agents and the systems they access, the platform evaluates agents in real time against security policies by considering each agent's identity, task, target resource, and requested action. The company is directly relevant to the news as it publicly launched with funding to expand its engineering team and advance development of this runtime enforcement platform. AI Agent Risks: AI agents can be properly authenticated and use approved tools yet still take unauthorized actions, creating a need for runtime control points that connect identity, task context, and policy. Industry Focus: Startups are building solutions to observe agent behavior, identify risks, and enforce policies in real time as AI agents increasingly operate software and access organizational systems.

Categories

techai_agentsaimachine_learningvirtuals
View Original Tweet