Companies are forced to route every user query through an additional large language model to prevent malicious attacks. This process requires significant and expensive Graphics Processing Units (GPUs), creating a hardware bottleneck. This forces companies to either incur huge cloud computing costs or selectively check only a small portion of their AI traffic, leaving a wide loophole for hackers.
To solve this problem, the startup Lasso Security raised $30 million in a Series A round, introducing a new approach to securing enterprise AI that eliminates the need for expensive GPUs.
To support the next generation of AI security systems, Lasso Security launched the LEAP product. This new product completely changes the mathematics of AI security because, instead of routing every query through a slow and expensive model, LEAP bypasses the GPU. LEAP runs directly on standard server processors, meaning on CPUs. The result is a transformer engine that delivers a security verdict in less than five milliseconds, which is thousands of times faster than existing market solutions.
Product Officer Ofir Dor noted that while the idea of using AI to protect other AI sounds impressive, it becomes unfeasible when calculating costs for millions of queries. LEAP solves this problem instantly.
When more nuanced judgment is required, a second engine—RAPID—is activated. Lasso has developed a highly intelligent routing layer. If the query is simple, LEAP processes it instantly. If it is complex, it is passed to RAPID, which assesses intricate queries against safety policies stated in plain language. The advantage of this two-tier system is its high economic efficiency: the vast majority of traffic never engages the GPU, ensuring robust protection without colossal computational costs.
Furthermore, Lasso has not just created a defensive barrier; they have also developed offensive weaponry. Their attack team constantly subjects client applications to automated 'red team' attacks. This is not just about basic constraint bypasses (jailbreaks); they conduct autonomous multi-step attacks. In one alarming real-world test with a global healthcare provider, the Lasso system was able to access patient records and actively change prescriptions. The discovery of such large-scale vulnerabilities makes their security software exceptionally effective.
This aggressive approach yields significant results: revenue has grown by 500% over the last year. Major players such as BMW, eToro, Fiverr, and the U.S. Department of Homeland Security have already joined their clientele.
This new $30 million funding round, led by ClearSky Advisors, increases Lasso's total funds to over $37 million. Existing investor Entrée Capital has doubled its initial seed investment. New investors, including iAngels, Singtel Innov8, Mindset, and Swish Data, also participated in the round. CEO Elad Schulman has a clear vision for how this influx of capital will be used.
Lasso is actively expanding its engineering base in both Tel Aviv and New York offices. In addition to hiring personnel, the company is actively entering federal and strictly regulated markets. Lasso Security is also increasing its sales reach across North America and Europe. For companies aiming to safely implement AI, the message is clear: you can now protect your AI agents without purchasing a single GPU.