Lakera

Real-time AI security guardrails against prompt injection and LLM threats

Hand-reviewed by the Toolglade team·Updated Aug 2026

ai-security#ai-security#prompt-injection#llm-guardrails#genai-security

Free plan Free trial Claimed API Self-hosted Teams

Toolglade’s take

Lakera is a leading name in the fast-emerging LLM security category, and its Gandalf game gives it a genuinely differentiated, ever-growing source of adversarial training data. The 2025 acquisition by Check Point (reported ~$300M) is a strong validation but also means buyers should watch how the product evolves inside a larger security portfolio. Pricing is not public beyond a free tier, so enterprise costs require a sales conversation. As with all guardrail tools, its detection claims should be validated against your own application's threat model rather than taken at face value.

About Lakera

Lakera provides real-time AI security guardrails via Lakera Guard, detecting prompt injection, jailbreaks, and data leakage, with threat intelligence from its Gandalf game.

Lakera Guard acts as a security layer for generative AI applications: every input and output passes through its detection engine before reaching or leaving the model, flagging or blocking threats such as prompt injection, jailbreak attempts, and data leakage. The company cites high detection rates with low latency and false-positive rates, positioning Guard as production-ready protection for LLM-powered products. A distinctive asset is Gandalf, Lakera's AI security game that has attracted over a million players and generated tens of millions of adversarial prompts. This crowdsourced red-teaming feeds Lakera's threat intelligence across 100+ languages, continuously improving detection as attack techniques evolve. Founded in Zurich in 2021 by engineers from Google, Meta, and aerospace, Lakera was acquired by Check Point in 2025 in a deal reported around $300 million, folding its AI security capabilities into a major cybersecurity vendor. It targets enterprises building LLM applications that need runtime guardrails and does not publish pricing, offering a free tier alongside custom enterprise plans.

TL;DR

Lakera provides real-time AI security guardrails (Lakera Guard) against prompt injection and LLM threats, fed by its Gandalf game and now owned by Check Point.

Company overview

Lakera is a Zurich-based AI security company founded in 2021 by engineers from Google, Meta, and aerospace. It builds runtime guardrails for generative AI applications and was acquired by Check Point in 2025.

The company is a prominent name in the emerging LLM security category, combining a production detection engine with crowdsourced threat intelligence from its Gandalf game.

Product features

Lakera Guard inspects every LLM input and output, detecting and blocking prompt injection, jailbreaks, and data leakage with high detection rates and low latency. It is API-first and covers 100+ languages.

Gandalf, its AI hacking game, has generated tens of millions of adversarial prompts from over a million players, continuously improving Lakera's detection as attack techniques evolve.

Target market

Lakera targets enterprises and AI product teams deploying LLM applications that require real-time security guardrails.

Buyer personas

End users

AI/LLM developers and AppSec engineers integrating guardrails.

Buyers

CISOs, heads of AI, and security engineering leaders.

Key influencers

Platform and ML engineering teams evaluating detection quality.

Ideal customer profile

An enterprise building customer-facing or high-risk LLM applications that needs runtime protection against prompt injection and data leakage.

Funding & performance

Lakera raised venture funding and was acquired by Check Point in 2025 in a deal reported around $300 million; verify details with the vendor or Crunchbase.