Lakera Guard
LLM security layer that blocks prompt injection data leaks and jailbreaks with a simple API policies dashboards and community to production tiers.
Overview
Lakera Guard protects generative AI apps in real time. You call a single API around prompts and responses and Lakera classifies risks like injection, exfiltration, secrets exposure or unsafe content before it reaches your model or user. Policies allow different guardrails per route so a public chatbot can run strict filters while an internal agent uses a tuned profile.
Central analytics show attack types and sources so you can harden flows and prove compliance. A free community plan helps teams prototype quickly then production tiers scale to higher request volumes with SSO, alerting, custom rules and SLAs. Documentation covers latency budgets, false positive tuning and red team methods so you can strike the right balance between safety and usability.
Lakera ships client libraries, supports major LLM providers and integrates with incident tools to keep your governance loop tight from detection to response.
Key features
- Single API call to detect injection leaks and jailbreaks
- Policies per application route to tailor risk tolerance
- Dashboards with attack analytics for compliance needs
- Low latency design to protect real time assistants
- Custom rules and allow lists for domain specifics
- SSO alerting and SLAs on paid production plans
- Client libraries and examples for fast adoption
- Works across providers to guard multi model stacks
Best for
- Protect a public chatbot from injection and jailbreak attempts
- Shield agents that browse tools and APIs from exfiltration
- Meet compliance by logging and reporting blocked risks
- Tune policies to reduce false positives in key paths
- Create allow lists for approved actions or domains
- Alert security teams with webhooks when threats spike
- Apply stricter filters on anonymous traffic at the edge
- Run red team simulations to validate coverage over time
Capabilities
Risk signals
Identify injection exfiltration jailbreak and unsafe content with a single protective API around requests and responses.
Policies per route
Apply stricter or looser rules by endpoint so public and internal apps follow the right guardrails.
Analytics and alerts
Track attack patterns false positives and volume then alert teams or tighten filters.
Providers and tools
Use SDKs and webhooks with your LLM stack and incident management to close the loop.
Frequently Asked Questions
How does pricing start for Lakera Guard?
A community plan is free to start and paid tiers begin around $99 per month with higher volumes and enterprise features available.
Will guardrails add latency to chats?
Lakera documents low overhead and options to parallelize checks to keep experiences responsive.
Can I tune rules for my domain?
Yes, you can add custom policies allow lists and thresholds for sensitive routes.
Does it work with multiple LLM providers?
Lakera supports popular providers so mixed stacks can be protected consistently.
How do I monitor what was blocked?
Dashboards and webhooks show attack types sources and actions for audits and response.
