Lakera Guard review
A runtime security layer for screening AI prompts, agent actions and outputs.
Reviewed by iTechGuides Editors · Editorial team · Updated Oct 2026
Lakera Guard is a runtime security layer for generative-AI applications and agents, suited to teams that need to inspect interactions before they reach a model or user. It screens prompts, reference content, tool calls, tool responses and outputs through an API, with detection for prompt injection, jailbreaks, sensitive-data leakage, harmful content, malicious links and off-policy agent behavior. The product is documented under the Check Point AI Guardrails name following Lakera's acquisition by Check Point. Its model-agnostic approach and cloud or self-hosted deployment options make it relevant to mid-market and enterprise teams building AI systems into products or workflows.
The breadth of screening is a central strength: controls cover both model-facing content and agent activity, including tool allow and deny lists and dangerous-deviation detection. Detect mode supports monitoring, while Enforce mode can block flagged interactions. Custom guardrails and configurable detection policies let teams adapt rules to their use cases; dashboards provide request logs, analytics and investigations, with SIEM export for security operations. Integrations include Amazon Bedrock, Google Cloud, Microsoft Copilot Studio, Salesforce Agentforce, n8n and Relevance AI. This makes Guard a fit for teams that need a gateway-style control layer across different agent environments rather than a filter focused only on final model output.
The published plans show a free Community option alongside Pro and Enterprise; the vendor directs prospects to contact its team for paid-plan details. That gives teams a way to start without a stated subscription cost, but buyers comparing paid tiers will need to discuss pricing directly. Email, community and documentation are listed as support channels. Choose Lakera Guard if runtime inspection, agent behavior controls and cloud or self-hosted deployment align with your security design. It is less suitable for teams looking for a standalone interface requiring no runtime integration, or for buyers who need transparent paid-tier amounts to evaluate options before contacting sales.
Lakera Guard pros and cons
- Where it wins
- Screens prompts, context, tool activity and outputs through an API
- Detects prompt attacks, sensitive-data leakage and unsafe content
- Cloud or self-hosted deployment with custom policies and investigations
- Where it doesn't
- Requires integration into an AI application's runtime flow
- Advanced policy configuration may take security-team effort
- Paid plan amounts are not published
Lakera Guard fact sheet, pricing and score →
Advertiser disclosure: iTechGuides is reader-supported. Vendors can pay for top positions in our rankings and for a place on other products' pages, and we may earn a commission when you click some links. How we rank.
Last updated · How we research and update