Lasso Security

Founded around language model security

Lasso Security was founded in 2023 in Tel Aviv by Elad Schulman, Ophir Dror, Lior Ziv and Yuval Abadi, according to Calcalist. According to the report, three of the four founders held key roles at Segasec, which was acquired by Mimecast in 2020. The new company focused on securing generative artificial intelligence and language models. In November 2023, Calcalist reported a $6 million seed round led by Entrée Capital with participation from Samsung Next. The current team page lists Schulman as CEO, Dror as CPO, Ziv as CTO and Abadi as COO, all as co-founders.

From usage discovery to runtime protection

The company's website describes a process that starts with discovering the artificial intelligence applications and agents in an organization and assessing their risk. Next come attack tests, and runtime protection checks the interactions with the system. The company stresses the context and the agent's intent, alongside the content of the requests and responses. The product is meant for organizations that run models and applications and want to set how users and agents access them. It connects visibility into usage with governance and protection rules in day-to-day operation. According to the website, the findings from attack tests feed back into the protection process.

Different engines for different cases

In September 2026 the company introduced LEAP, a protection engine that is not based on a transformer and runs on ordinary processors without a graphics accelerator. Alongside it runs RAPID, a local language model that acts as a judge for cases that need complex judgment against a policy written in plain language. A routing layer sends most of the traffic to LEAP for inspection, and the suitable cases to the deeper inspection. According to the company, the goal is to reduce the use of expensive hardware for every request, while keeping a path for complex cases. So running without a graphics accelerator refers to the LEAP engine, not to every component of the platform.

New funding and use in organizations

On 2 September 2026, Lasso Security announced a $30 million round led by ClearSky. Entrée Capital, iAngels, Singtel Innov8, Mindset and Swish Data also took part. The company said the funding is meant to expand engineering and its business with regulated organizations and in the US federal market. The announcement listed BMW, eToro and Fiverr among its customers, along with the US Department of Homeland Security. The announcement also includes a quote from the chief information security officer of Kaufman Rossin about visibility into artificial intelligence use and support for data loss prevention. The customer names are given in the company's announcement and in the SiliconANGLE article that accompanied it.

Attack testing and the limits of comparison

According to the announcement, the company's red team tests applications and agents, from single requests to attack sequences that unfold over a conversation. The findings are used to improve the product. SiliconANGLE noted that the speed and accuracy tests behind the LEAP announcement are Lasso's own tests. Alongside the fast engine, the platform keeps a path for complex judgment through RAPID. This is the protection structure the company presents: ongoing inspection of requests and actions, with handoff to another engine when further judgment is needed. The distinction between the engines is part of the stated design, alongside the attack testing and monitoring.