Horizon3.ai Advances Autonomous Cyber Defense with Predictable AI Architecture

Event summary

  • Horizon3.ai's new research addresses unpredictability in autonomous AI cyber defense, making it safe for real-world deployment.
  • The tool-mediated architecture ensures stability by constraining AI actions to a finite, pre-approved catalog.
  • Tested on 161 organizations across 25 industries, the system reduced attacker success by 59% and eliminated off-catalog hallucinations.
  • The research demonstrates model-agnostic safety, with both advanced and smaller AI models operating within the constrained action space.
  • The NodeZero® platform can now autonomously adjust EDR policies, including Microsoft Defender, without degrading defensive posture.

The big picture

Horizon3.ai's breakthrough challenges the assumption that only the most advanced AI models can deliver reliable cybersecurity results. By making autonomous defense predictable and controllable, the company is positioning itself at the forefront of a shift towards AI-driven, real-time cybersecurity adjustments. This development aligns with the broader industry trend of integrating AI into critical infrastructure, potentially setting a new standard for cyber resilience.

What we're watching

Adoption Pace
How quickly organizations will integrate autonomous AI cyber defense into their live environments.
Model Flexibility
Whether smaller, cost-efficient AI models can sustain the same level of safety and effectiveness as advanced models.
Regulatory Impact
The pace at which regulatory frameworks will adapt to the deployment of autonomous AI in cybersecurity.

Related topics

Cybersecurity