Lasso Security Now Integrates with TrueFoundry AI Gateway

Every AI security problem looks manageable when you're running one application on one model. The real challenge appears at scale, when teams are building on different models and shipping at a pace that makes per-application security review impractical. 

‍

Lasso exists for this moment, when scale outpaces per-application review. And today, we're announcing our integration with TrueFoundry AI Gateway. This brings Lasso's security layer directly into the central control plane that enterprise teams already use to manage their AI infrastructure.

‍

Why the Gateway Is the Right Place for Security

TrueFoundry AI Gateway is how enterprise teams unify access to large language models across providers and environments. Instead of every application maintaining its own model connections and API keys, teams point everything at the Gateway and let it handle routing, fallback, and observability.

‍

That architecture gives teams control over how traffic flows. It doesn't, on its own, inspect what that traffic contains. Every prompt sent to a model and every response that comes back is a potential vector for data leakage, policy violation, or adversarial manipulation. A gateway built for routing can see that traffic moved without being able to see what it said.

‍

That's where Lasso operates in this integration.

‍

What Lasso Adds at the Gateway Layer

Lasso is purpose-built for the way large language models behave in production. Rather than relying on keyword filters, Lasso uses intent classification to understand what a prompt is actually trying to do. It catches adversarial inputs, prompt injection attempts, and jailbreak patterns that generic filters miss.

‍

Lasso now acts as a guardrail configured directly in the TrueFoundry dashboard. Every request that comes through the Gateway is evaluated by Lasso before it reaches a model. If Lasso returns a block verdict, the Gateway halts the request and returns a blocked response. If a policy calls for masking PII, the redaction happens before the prompt is forwarded, and the response the user sees is already clean.

‍

Responses can also be scanned on the way back out, so harmful or sensitive content in model outputs is caught before it reaches the user.

‍

This security architecture covers the full round trip, applied consistently across every model and every application that runs through the Gateway.

‍

One Configuration, Every Application

Configuring Lasso at the Gateway level means security policy is managed in one place. Teams building new applications don't need to instrument Lasso separately; the coverage is already there. And because Lasso's findings are paired with TrueFoundry's request tracking, every blocked prompt, masked field, and policy decision is captured in a complete audit trail.

‍

For security teams, that means AI activity across the organization becomes visible and auditable without requiring per-team coordination or custom integrations.

‍

Start Securing Your AI Today 

The integration is available now. To enable it, add Lasso as a guardrail in your TrueFoundry account, point it at your deployed Lasso endpoint, supply your API key, and attach it to the models you want to protect.

For setup documentation, visit the TrueFoundry integration reference or reach out to the Lasso team directly.