Lasso Security
A security platform that tests and monitors company AI systems for attacks like prompt injection, flagging and blocking manipulation attempts before they cause damage.
🔗 Visit Lasso Security
Description
AI chatbots and agents can be tricked by cleverly worded text into ignoring their instructions and doing something they shouldn't — a trick called "prompt injection," roughly the AI equivalent of a con artist talking their way past a guard. Lasso Security builds tools that actively try to trick a company's own AI systems the way an attacker would, find the weak spots, and then watch production traffic in real time to catch and block real attack attempts as they happen.
Lasso Security is an enterprise AI security platform covering discovery, risk management, automated adversarial (red-team style) testing, and runtime protection for autonomous agent systems. Its headline capability, Intent Deputy, claims prompt-injection detection in under 50ms at 99.83% accuracy. It also covers MCP security with real-time threat visibility. Pricing isn't public — enterprise sales cycle with a custom quote based on deployment and usage, targeting CISOs and security leaders in regulated industries (finance, healthcare, public sector) rolling out generative AI at scale.
💬 Our review
The short version: if you're deploying generative AI broadly across a regulated organization and need both proactive testing (finding weaknesses before attackers do) and runtime protection (catching attacks live), Lasso covers both ends rather than just one.
Lasso's differentiator against narrower point solutions is breadth — discovery, adversarial testing, and runtime enforcement in one platform, versus competitors that specialize in just one layer (e.g., SolonGate's MCP proxy or Entro's identity focus). The claimed 50ms/99.83% accuracy on prompt-injection detection is a strong number if it holds in your specific environment, but it's a vendor-reported figure with no public independent benchmark, so validate it against your own attack patterns during a POC rather than taking it at face value. As with peers in this category, pricing is fully opaque, and the target customer is explicitly regulated-industry enterprises — this isn't built or priced for a small team's occasional AI security check.
💰 Pricing
📊 Global score
🤖 AI-enriched data
Tarification non publique, devis personnalisé selon déploiement et usage
Pros
Couvre à la fois le test proactif (red-team automatisé) et la protection en temps réel
Détection d'injection de prompt annoncée sous 50ms à 99,83% de précision
Visibilité complète sur la sécurité MCP
Ciblage clair des secteurs réglementés à fort enjeu
Cons
Chiffres de précision annoncés par le fournisseur, sans benchmark indépendant public
Tarification totalement opaque
Cycle de vente entreprise, pas d'auto-service pour petites structures