Lakera

AI security software from Check Point for protecting employee AI usage, discovering agents, enforcing runtime guardrails, and testing applications against adversarial attacks.

About Lakera

Lakera, a Check Point company, provides security capabilities for employee AI usage, AI applications, and autonomous agents. Its offerings cover workforce visibility, agent discovery, runtime protection, and adversarial testing. These capabilities help organizations examine where AI is used and apply controls around sensitive information and connected tools.

Current technical documentation uses the names Check Point AI Agent Security and AI Guardrails. AI Agent Security combines discovery, risk assessment, and runtime protection. AI Guardrails is also available as a standalone runtime layer for teams integrating security screening into their own applications or gateways.

Workforce AI Security addresses employee interactions across browsers, desktop applications, developer tools, and connected services. AI Red Teaming evaluates application-specific weaknesses and checks whether model or system changes introduce new security failures.

Key Features

Workforce AI visibility: Discovers employee AI usage and supports policies intended to reduce sensitive-data exposure.
Agent discovery: Builds an inventory of agents, their tools, and connected MCP servers across supported platforms.
Agent risk assessment: Evaluates agent configurations and explains contributing risk factors.
Runtime guardrails: Screens interactions for prompt attacks, data leakage, content violations, and behavior outside configured policies.
Tool-use controls: Supports restrictions on which tools agents can call and screening of tool-related content.
AI red teaming: Tests application defenses against adversarial inputs, unauthorized actions, and regression risks.

Pricing

Free account: Free account access is available for getting started with AI Guardrails. Verified usage allowances were not established.
Enterprise offerings: Contact sales. Published product pages do not specify paid amounts or billing commitments.

Use Cases

Apply data-protection policies to employee use of AI applications.
Discover agents and assess the tools or systems they can access.
Screen prompts, responses, and tool interactions within custom AI applications.
Test deployed applications after changes to models, instructions, or connected capabilities.

Pros & Cons

Pros:

Covers workforce usage, application protection, and agent-security workflows.
Offers a standalone guardrails layer as well as broader agent-security capabilities.
Supports SaaS and self-hosted deployment for AI Guardrails.

Cons:

Audio Defense is described as Early Access.
Image-threat protection is described as coming soon.

Integrations

Amazon Bedrock, Amazon Bedrock AgentCore, Google Cloud, Microsoft Copilot Studio, Salesforce Agentforce, n8n, Relevance AI

FAQ

What is Lakera?

Lakera is a Check Point company providing security capabilities for employee AI usage, AI applications, and agents.

How do AI Agent Security and AI Guardrails differ?

AI Agent Security includes discovery, risk assessment, and runtime protection. AI Guardrails provides the runtime screening layer as a standalone offering.

Which threats can AI Guardrails screen for?

Documented controls address prompt attacks, sensitive-data leakage, content violations, malicious links, and agent behavior outside configured policies.