Lean SuperIntelligence Introduces Paired Cybersecurity Models

Lean SuperIntelligence closed an initial financing round and introduced two cybersecurity language models, the company said in a press release. The financing amount and investor names were not disclosed.

LSI-Offense-1 discovers, exploits, and validates software vulnerabilities, while LSI-Defense-1 detects threats, investigates alerts, and recommends responses. The models train through the company's Cyber World Models system, where attacks found by the offense model become training material for the defense model, and successful defenses create new challenges for the offense model.

The models support air gapped, on premises, sovereign, and managed deployments using customer data. LSI says both models outperformed general purpose models on public and independent security benchmarks, but it did not publish detailed results in the announcement.

LSI-Offense-1 and LSI-Defense-1 are available to a limited group of security vendors, managed security service providers, and large enterprises.