Artemis Security launches Orion-1 AI for cybersecurity

Artemis Security, a US-based startup, has launched Orion-1, a specialized artificial intelligence model designed to autonomously investigate and respond to cyberattacks on corporate networks. The model aims to address a growing issue in cybersecurity where general-purpose AI models are often restricted by “guardrails” to prevent misuse, which can inadvertently hinder defensive operations.

Orion-1 analyzes log data from authentication services, cloud infrastructure, corporate devices, SaaS applications, and network equipment. By linking disparate signals, the model can reconstruct the progression of an attack and initiate countermeasures. In the company’s proprietary “Decision-Grade Readiness” (DGR) benchmark, Orion-1 achieved an attack reconstruction score of 80.8, significantly outperforming general models such as Claude Opus 5.5 and GPT-6 Sol, which scored 58.3.

While Artemis describes Orion-1 as a Frontier Foundation Model, the company has primarily detailed its specialized training on security-specific tasks, such as identifying suspicious log patterns and writing detection rules. It remains unclear whether the model is built upon an existing foundation from another provider.